Altium REJECTS takeover bid from Autodesk

[EEVblog]

Altium just rejected a takeover bid from Autodesk.
It was in talks with them, but considered the unsolicited offer that was 30% above the market price was undervalued. Their shared just jumped 33% as a result.

[nigelwright7557]

Given you can get perfectly usable software for free I am surprised anyone buys it.

[Monkeh]

Altium just rejected a takeover bid from Autodesk.
It was in talks with them, but considered the unsolicited offer that was 30% above the market price was undervalued. Their shared just jumped 33% as a result.

Probably one of the smartest decisions they've ever made.

Given you can get perfectly usable software for free I am surprised anyone buys it.

Oh give over.

Anyway, I thought KiCad was useless?

[KE5FX]

WTF.  Hosing EAGLE wasn't enough for those clowns?!

[WattsThat]

Every time I see one of these (attempted) mega-mergers, I’m reminded of a corporate entity name from an old Mel Brooks movie:

Engulf and Devour

[ataradov]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

[T3sl4co1l]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

Guessing it's the same thing that apparently makes it easy to draw polygons everywhere.  Think I had asked before but either didn't get a conclusive answer, or since forgot it...

Tim

[EEVblog]

Probably one of the smartest decisions they've ever made.

If the market tanks they might come to regret it from a finanical point of view.
Note they didn't reject the buyout on principle, it's just that they weren't offering enough. Stay tuned...

[EEVblog]

Every time I see one of these (attempted) mega-mergers

It's wasn't a merger attempt

[RiZsho]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

Maybe Zuken? It's a Japanese company.

[delfinom]

The sad thing is Autodesk is "niche" to the world so they won't get the anti-trust beating they deserve

[EEVblog]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

Maybe Zuken? It's a Japanese company.

Likely, Zuken is very big.
Bigger than Altium, and was #2 at one point, but that was a decade ago.
They always featured in Altium anual reports as a main competitor they were trying to chase.

[Bud]

This may explain Altium's aggressive tactic to onboard as many subscription renewals as possible. I lost count of the "Last and Final " offers they sent over the last year.

[delfinom]

This may explain Altium's aggressive tactic to onboard as many subscription renewals as possible. I lost count of the "Last and Final " offers they sent over the last year.

Nah, that's just altium's army of commission based sales people needing to make money and just following whatever used car salesman technique they know

[EEVblog]

This may explain Altium's aggressive tactic to onboard as many subscription renewals as possible. I lost count of the "Last and Final " offers they sent over the last year.

Nah, that's just altium's army of commission based sales people needing to make money and just following whatever used car salesman technique they know

Yep, this 

[ajawamnet]

One of the best things that ever happened to KiCAD was Adesk buying Eagle. There's so much talk on the Altium Forum about how a lot of seasoned users are thinking about having their companies take the Altium subscription money and donate it to KiCAD.  Altium recently mentioned how their code has bloated out.  They'd never have the financial mass to re-write it from scratch and get rid of stuff like FORTRAN 77 - yep it actually has that and a bunch of other silly running under the hood.   If I were Altium I'd do the same thing - sell it while it's worth something. Looks like our fearless forum leader was wise in getting the hell out of there.   F'ing sad what it turned into at the expense of corporate greed. Really sad...

[ajawamnet]

Post I did on the Altium Forum

And again I will mention that Autodesk's subscription based model (Ie SaaS or rental) was also a response to their previous perpetual license software model when it was almost dealt a significant blow in the initial Vernor vs. Autodesk.  These CAD tools blew their wad in the late 1990's giving most users all the tools they needed to replace drafting boards and tape up.   And as I've posted before -    shows that SW 1995 is is just as capable for most designs as is the latest versions.   

This happened in the DAW industry which started in the late 1980s - well, if you follow my links -
http://www.ajawamnet.com/ajawamnet/a_definite_non_turd_Roger_Nichols.html
it started  earlier than that with Roger Nichols (six-time Grammy winning engineer that did Steely Dan and others ) and the Wendel in the mid-70's,

By the early 2000's these music recording tools gave you close to everything you needed to replace a large, expensive recording studio and the limitations of editing multitrack tape with razor blades.  Just like word processors replaced typewriters.

The key is easy editing as compared to the manual methods from decades ago.

Eventually, suites like  Reaper - which was done for the right reasons  -  became on par with software such as Protools (Profools). Listen to Justin from Reaper  talk about it: https://youtu.be/vfaQrOeb_F0?t=202    and at 3:50:

"... when I left [AOL after they bought Winamp], I came away from it wanting to avoid that in the future; wanting to just make things for the sake of making them .. and not have to constantly justify everything with business decisions/motivations. The ability to just make software for the purpose of making it... for the end goal of making something that's really powerful and enjoyable to use"

So, in that realm - just like it is in CAD and other software tools - the DAW doesn't matter;  it's the people's whose hands it's in - that's what makes hit recordings. 

Stephen King wrote Carrie on his wife's portable typewriter.

Just like these DAW tools, people like Justin that are in a financial position to do some really great coding for the right reasons will supplant lower - to mid level commercially driven products.  Look at things like Open Office and such. 

It comes down to "What do you really need to do what you want?"

As to Altium's value - yea, I'd grab the billions from Adesk and run. Altium is in what I would evaluate as a conundrum since it isn't big enough to compete with large, IC design tool companies that also offer products in the same space, and on the other side competing with the ever-escalating Freeware capabilities.   Companies are willing to spend ungodly amounts on IC design tools since the fabrication costs of first articles is stoopid expensive. So these companies have the financial mass to do things like re-write and maintain

Autodesk buying Altium is like humans that hook up with some other, at-the-time desirable human, only to discard them in the end when they tire of them  and go looking for another.  They fail to get the fact that there's more to a fulfilling life than cursory experiences ( I say this being an ugly old fool so take it for what it's worth).

After four decades of doing this it's my opinion that software in the hands of talentless people will be limited to just that.. software. Just like most digital hardware is just a collection of gates and such without software, software without knowledge and care is just as useless if not dangerous.

[ajawamnet]

And one of the long time users posted this:

xxxxxxx wrote:
.., since you use their IP to make your IP all yourz IP is ourz...lol. I don't know what marketing genius at Autodesk thought that would go over well.

my response is:
This will eventually be legally tested,   I'm sure of it

[Mark19960]

Autodesk. Meh.

Some people will stick with Altium or "AutoDesk Altium" as I am sure they would rename it if they get their claws on it. 
IMHO, Altium is better off as they are rather than letting the goons at Autodesk have a go.

[floobydust]

Siemens bought Mentor Graphics for $4.5 billion in 2017 "to expand its industrial software portfolio" and they are close with TSMC with their tools.

It's usual MBA mentality to scoop up CAD companies as one big blog, leaving us with disjointed, neglected, redundant tools.
The usual mess nobody knows how to manage or look after. All in the name of what? Megalomania?

Autodesk would have EagleCAD and Altium, OK now what? Kill off one of them, merge them into a dog's breakfast mega CAD program, just keep selling the two but with a massive price hike and forced cloud use, or just waste everyone's time rebranding and logo changing it. It's so stupid looking at it from 20,000 feet.

[ajb]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier. 

Post I did on the Altium Forum

And again I will mention that Autodesk's subscription based model (Ie SaaS or rental) was also a response to their previous perpetual license software model when it was almost dealt a significant blow in the initial Vernor vs. Autodesk.  These CAD tools blew their wad in the late 1990's giving most users all the tools they needed to replace drafting boards and tape up.   And as I've posted before -  <link snipped to remove inline>  shows that SW 1995 is is just as capable for most designs as is the latest versions.   

"What you can make" is really only one aspect of the value proposition of any piece of design software.  The other crucial part of it is "How fast/easy is it to make something", which depends on a lot of factors including how sophisticated the user-accessible tools are, how easy it is to understand those tools, how fast and consistent the user interface is, how fast the underlying computation is, and how reliable the whole lot is.  In that light, the video you posted shows that SolidWorks has come quite a long way between 1995 and 2012, which is the point it seems like the narrator was trying to make. 

Same thing with AutoCAD, which, sure, is still a 2D drafting tool and still uses some of the same commands and interface language that it did 25 years ago or whatever, but there have been a ton of improvements since then in the sophistication of the tools, the speed of the user interface, and overall capabilities.  In either case, is that enough to make everyone want to pay for annual updates? No, hence the subscription model push, but there is at least forward progress being made with both SW and AutoCAD. 

In the case of Altium, there is also progress being made, and in many ways the software is definitely better in later versions than in older versions.  But compared to a lot of other professional software, Altium still has a LOT of room for fundamental improvements and their trajectory seems to be less consistently positive in terms of software quality.  There's still a bunch of inconsistency and redundancy in the user interface, outright bugs that have been there forever, regressions are common (how tf did they make interactive routing WORSE?! It's a PCB design tool so that's kind of important?!), and the improvements we see tend to not be in the areas that most need it. 

So it seems to me that what makes Altium's subscription model unattractive isn't that the software is currently good enough, but that the updates aren't making it tangibly better to a degree that justifies the price.  I would be much less unhappy to pay $2k/yr if it consistently got me better software.  Instead I'm being asked to pay $2k/yr to hope that maybe they'll actually fix a couple of bugs that I care about or add a legitimately useful feature and that nothing serious will break.  Now they're offering their 365 cloud thing into the bargain, which I'm sure is legitimately attractive to a lot of people, and I guess expecting that to drive subscription sales, which is fine and might even be successful for them (they seem to think it will be, since they think they were undervalued at +40%), but it does make one even less hopeful about how concerned they are with improving the fundamentals of the design software. 

[nigelwright7557]

Oh give over.

Anyway, I thought KiCad was useless?

Kicad is wonderful, it gave me the push to forget PCBCAD software and move on to something more profitable that cant be given away.
Thank you Kicad.....

[olkipukki]

Altium just rejected a takeover bid from Autodesk.
It was in talks with them, but considered the unsolicited offer that was 30% above the market price was undervalued. Their shared just jumped 33% as a result.

... and the price back to late 2020 level 


Autodesk make sense to acquire Altium, they already have low cost all inclusive Fusion 360 package, but missed Inventor-style product.
What else they can do and when last time they build anything from scratch? The cash is cheap these days, time - very expensive. I bet they will buy Altium sooner or later

[olkipukki]

Autodesk would have EagleCAD and Altium, OK now what?

Can you still buy Eagle? I doubt, this is part of Fusion 360.

I guess Altium would be same as Inventor-price category or part of 'budget' collection

[EEVblog]

One of the best things that ever happened to KiCAD was Adesk buying Eagle.

Yep!

Altium recently mentioned how their code has bloated out.  They'd never have the financial mass to re-write it from scratch and get rid of stuff like FORTRAN 77 - yep it actually has that and a bunch of other silly running under the hood. 

Where did you hear that?
Protel for DOS was written from scratch by Nick Martin in Borland Pascal, and then moved to Delphi once they moved to Windows. It never used Fortran.

[EEVblog]

Autodesk make sense to acquire Altium, they already have low cost all inclusive Fusion 360 package, but missed Inventor-style product.
What else they can do and when last time they build anything from scratch? The cash is cheap these days, time - very expensive. I bet they will buy Altium sooner or later

Wait for the share market cash and then that offer will sound like a real good deal 
Altium is at a massively high P/E ratio.

[EEVblog]

So it seems to me that what makes Altium's subscription model unattractive isn't that the software is currently good enough, but that the updates aren't making it tangibly better to a degree that justifies the price.  I would be much less unhappy to pay $2k/yr if it consistently got me better software.  Instead I'm being asked to pay $2k/yr to hope that maybe they'll actually fix a couple of bugs that I care about or add a legitimately useful feature and that nothing serious will break.  Now they're offering their 365 cloud thing into the bargain, which I'm sure is legitimately attractive to a lot of people, and I guess expecting that to drive subscription sales, which is fine and might even be successful for them (they seem to think it will be, since they think they were undervalued at +40%), but it does make one even less hopeful about how concerned they are with improving the fundamentals of the design software.

Altium spent over a decade trying to get users to upgrade from Protel 99SE, especially in China where they didn't give a rats arse about any latest features, they just stuck with the same tool that worked. It was still a major sales hurdle when I left in 2011.
It didn't help that the mid to late 2000's ushered in the destined to fail focus on FPGA's at the expense of the PCB tool.

[EEVblog]

https://pcdandf.com/pcdesign/index.php/editorial/menu-news/design-news/15777-autodesk-makes-4b-bid-for-altium

[tszaboo]

Well, this is just the usual story.
1. Company wants a PCB design software
2. Company buys the cheapest possible, Eagle
3. Eagle is trash, impossible to make good designs in it
4. Spend years and countless engineers quitting and failed designs, sunken cost fallacy, cognitive dissonance
5. They finally ask the engineers what they want
6. Company buys Altium designer
7. PCBs are made, they are nice and they work

We all seen this happen. Its 100% the same story.

[Batang]

I remember those halcyon days of doing PCB layouts using BISHOP GRAPHICS materials.

Must be getting old.

[Yansi]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

Zuken

//okay, somebody was faster than me.

//I use Zuken almost daily at work now, but I wouldn't say it is anywhere close to being a user friendly tool. No way in hell.

[coppice]

Given you can get perfectly usable software for free I am surprised anyone buys it.

Really? You know a free package that's efficient to use for large scale board development?

[coppice]

Imagine autodesk acquires and kills another industry leading product.  I bet industry would not be too happy.

At the same time, I see a lot of schematics come from Japan prepared in something that is clearly not Altium. Does anybody knows what do they typically use. It looks like a very common piece of software, since all schematics look the same.

Zuken

//okay, somebody was faster than me.

//I use Zuken almost daily at work now, but I wouldn't say it is anywhere close to being a user friendly tool. No way in hell.

Zuken has a long history, but it used to be pretty much a Japan only product. Now I see their web site claims a lot of major western users. I wonder just how widely used it is outside Japan?

[coppice]

Well, this is just the usual story.
1. Company wants a PCB design software
2. Company buys the cheapest possible, Eagle
3. Eagle is trash, impossible to make good designs in it
4. Spend years and countless engineers quitting and failed designs, sunken cost fallacy, cognitive dissonance
5. They finally ask the engineers what they want
6. Company buys Altium designer
7. PCBs are made, they are nice and they work

We all seen this happen. Its 100% the same story.

I can't imagine why anyone buys Eagle. We used to find it extremely useful for designing evaluation and other similar small boards, with Eagle as a free package. You can pass your CAD files around, and anyone can immediatly work with them without needing to buy anything. However, as soon as it costs even $1 it has lost that benefit, and it has little to offer as a commercial package.

[Bud]

Given you can get perfectly usable software for free I am surprised anyone buys it.

Really? You know a free package that's efficient to use for large scale board development?

Agreed. It is totally different story when you get to work at Enterprise level.

[Monkeh]

Given you can get perfectly usable software for free I am surprised anyone buys it.

Really? You know a free package that's efficient to use for large scale board development?

He doesn't know what large scale development is.

[ataradov]

im just watching a random video about zuken, dont know anything about them. it looks good.

As a person having to review schematics from that tool, the only thing I observe about it is how annoying its PDF outputs are. They are just vector graphics, even for fonts, so i makes then unsearchable. They could have at least put hidden the under text.

I don't know if this is a setting or just a missing feature, but all the schematics I've seen were like this.

[free_electron]

a couple of weeks ago i was diggin aroudn to see what our far east collegues use for design tools ..

Zuken is one of them but there are much more other tools out there ...

Quadcept
Cadvance Alpha -> this was bought last year by zuken and killed.
Cadlus
OPuser
FIRST

www.cadlus.com
www.first-cad.com

Japanese schematics and boards have a very distinct layout and drawing style. just look at service manuals from onkyo , denon , sony , clarion. very different drawing and layout style.

[free_electron]

please not Autopest. Makers of such fine software like AutoCrap... and inventors of the drawing format from hell : DXF: Deficient exchange format. What drunken idiot came up with a drawing description format that does not have a scale embedded ?
Here is a drawing , but we can't be bothered to tell you if it is in inches, meters , furlongs or smoots.

Come on Dassault Systems, buy Altium. then we'll have real ecad/mcad synergy. Catia, Solidworks, Altium

[olkipukki]

Come on Dassault Systems, buy Altium.

Let's it have to be PTC for more fun 

[rsjsouza]

With such acquisitions so close to each other, I can't help but wonder that Autodesk hired bad consultants that gave them the wrong idea about Eagle, believing they would get an immediate share in the enterprise EDA market - instead, they got a foot on the mass market that is being shrunk by Kicad.

That or they may well go ahead, up the ante anyways and release a new subscription-based Altium. As for Eagle, either they kill it altogether (based on their financials) or it becomes an Altium-Lite package.

Oh well... Time will tell.

[thm_w]

With such acquisitions so close to each other, I can't help but wonder that Autodesk hired bad consultants that gave them the wrong idea about Eagle, believing they would get an immediate share in the enterprise EDA market - instead, they got a foot on the mass market that is being shrunk by Kicad.

That or they may well go ahead, up the ante anyways and release a new subscription-based Altium. As for Eagle, either they kill it altogether (based on their financials) or it becomes an Altium-Lite package.

Oh well... Time will tell.

They paid pennies for Cadsoft, maybe $25 million. The fact that they can add it in as a marketing bullet point in their yearly subscription package seems easily worth it to me.

I don't believe anyone would think that a company that cheap gives them significant market share.

[EEVblog]

They paid pennies for Cadsoft, maybe $25 million. The fact that they can add it in as a marketing bullet point in their yearly subscription package seems easily worth it to me.
I don't believe anyone would think that a company that cheap gives them significant market share.

Yes, it's all about adding an infographic segment in their anual report.
If they bought Altium then they would have two mid and free hobby level PCB tools (Circuit Studio + CircuitMaker), surely they would have to discontinue one, or at least stop development on it.
But then again, it doesn't take many programmers to keep a product alive, even a mainstream one. You'd be shocked at the (small) number of programmers Altium had working on the PCB tool.

[EEVblog]

I can't imagine why anyone buys Eagle. We used to find it extremely useful for designing evaluation and other similar small boards, with Eagle as a free package. You can pass your CAD files around, and anyone can immediatly work with them without needing to buy anything. However, as soon as it costs even $1 it has lost that benefit, and it has little to offer as a commercial package.

Eagle had the traction it did because the maker scene started using it. Adafruit, Sparkfun et.al because there was that free version that allowed you to do small boards which is what the entire maker arduino scene was all about.
Harly anyone in industry used it for serious high end work.

[PlainName]

https://pcdandf.com/pcdesign/index.php/editorial/menu-news/design-news/15777-autodesk-makes-4b-bid-for-altium

"Autodesk said it would finance the transaction using cash on hand and debt financing."

Isn't that - debt financing - where your target takes on the loan you got to buy it, so the target basically pays you to take it over? And then either has to raise it's prices to pay it off, or goes bankrupt and you pay a pittance for the assets from the liquidator. Either way, you get the company at very little real cost.

[Quick google later]
Seems I'm thinking of 'debt loading' which is unethical. 

[Analog]

A Dassault deal would make sense. Or Siemens could buy to gain market share. Pads could use an update and Altium could use something like Hyperlynx.

[ajb]

You'd be shocked at the (small) number of programmers Altium had working on the PCB tool.

No I wouldn't, I've seen the dismal pace of improvements and--even worse--number of regressions over several years of using it.  They clearly aren't putting enough resources into it, whether that's just a lack of developers or lack of testing. 

[nctnico]

I can't imagine why anyone buys Eagle. We used to find it extremely useful for designing evaluation and other similar small boards, with Eagle as a free package. You can pass your CAD files around, and anyone can immediatly work with them without needing to buy anything. However, as soon as it costs even $1 it has lost that benefit, and it has little to offer as a commercial package.

Eagle had the traction it did because the maker scene started using it. Adafruit, Sparkfun et.al because there was that free version that allowed you to do small boards which is what the entire maker arduino scene was all about.
Harly anyone in industry used it for serious high end work.

The latter goes for Altium as well. Every SoC design example which has some high speed interfaces and / or memory on it is made using Cadence Allegro. My guess is that KiCad is starting to eat into Altium's revenues as well and Altium can't climb up the ladder fast enough. If Autodesk transforms Altium into a subscribtion only model then it is probably the last nail in the coffin. The only thing currently missing from KiCad is a real, database driven part management system but that is on the development roadmap for the next version. Once that is there, there is very little reason to choose Altium.

[coppice]

I can't imagine why anyone buys Eagle. We used to find it extremely useful for designing evaluation and other similar small boards, with Eagle as a free package. You can pass your CAD files around, and anyone can immediatly work with them without needing to buy anything. However, as soon as it costs even $1 it has lost that benefit, and it has little to offer as a commercial package.

Eagle had the traction it did because the maker scene started using it. Adafruit, Sparkfun et.al because there was that free version that allowed you to do small boards which is what the entire maker arduino scene was all about.
Harly anyone in industry used it for serious high end work.

The latter goes for Altium as well. Every SoC design example which has some high speed interfaces and / or memory on it is made using Cadence Allegro. My guess is that KiCad is starting to eat into Altium's revenues as well and Altium can't climb up the ladder fast enough. If Autodesk transforms Altium into a subscribtion only model then it is probably the last nail in the coffin. The only thing currently missing from KiCad is a real, database driven part management system but that is on the development roadmap for the next version. Once that is there, there is very little reason to choose Altium.

I've seen quite a few big organisations dump Allegro in the last few years, and use Altium across their entire organisation. I'm not sure of the reasons, but Altium is being used for all kinds of exotic designs.

[PlainName]

Perhaps a fair number of Altium users aren't doing chip-level designs, so the lack of that very high-level stuff isn't an issue. After all, you wouldn't use Pagemaker to run off a letter, or a 747 to nip down the shops.

[nctnico]

I can't imagine why anyone buys Eagle. We used to find it extremely useful for designing evaluation and other similar small boards, with Eagle as a free package. You can pass your CAD files around, and anyone can immediatly work with them without needing to buy anything. However, as soon as it costs even $1 it has lost that benefit, and it has little to offer as a commercial package.

Eagle had the traction it did because the maker scene started using it. Adafruit, Sparkfun et.al because there was that free version that allowed you to do small boards which is what the entire maker arduino scene was all about.
Harly anyone in industry used it for serious high end work.

The latter goes for Altium as well. Every SoC design example which has some high speed interfaces and / or memory on it is made using Cadence Allegro. My guess is that KiCad is starting to eat into Altium's revenues as well and Altium can't climb up the ladder fast enough. If Autodesk transforms Altium into a subscribtion only model then it is probably the last nail in the coffin. The only thing currently missing from KiCad is a real, database driven part management system but that is on the development roadmap for the next version. Once that is there, there is very little reason to choose Altium.

I've seen quite a few big organisations dump Allegro in the last few years, and use Altium across their entire organisation. I'm not sure of the reasons, but Altium is being used for all kinds of exotic designs.

That is not the point. The question is: what is the added value of Altium when (not if) KiCad offers similar functionality? You can get professional support for KiCad if you want so that is not an issue. Which in turn raises the question on how sensible it is for Autodesk to acquire Altium. Allegro is a step up from Altium so in turn not likely to be threatened by KiCad however many companies won't need stuff like impedance and crosstalk simulations. The latter is unlikely to ever be implemented in KiCad because it is not trivial to do.

[EEVblog]

That is not the point. The question is: what is the added value of Altium when (not if) KiCad offers similar functionality?

KiCad might appear to offer similar functionaly to Altium if all you ever do is fairly basic to mid level boards. But if you do very complex mid to higher end stuff, Altium is still streets ahead.

[bson]

That is not the point. The question is: what is the added value of Altium when (not if) KiCad offers similar functionality?

KiCad might appear to offer similar functionaly to Altium if all you ever do is fairly basic to mid level boards. But if you do very complex mid to higher end stuff, Altium is still streets ahead.

But the vast majority of PCB design is extremely mundane, intentionally eschewing complexity and moving as much of that as possible to firmware.  If 90% of Altium licensees don't see a benefit, what will this do to Altium's revenue?  And this is only going to go in one direction, 90% may be 92% next year.  95% the year after that.  Eventually there isn't enough for Altium to sustain life.  The ten customers left who need some specific feature will just have to go find something else.

[james_s]

I'm a big fan of KiCad but I think Altium has a bigger lead than you give them credit for. Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon. A large amount of PCB design is not high end, but if you're a company that does 90% simple stuff and 10% stuff that requires a product like Altium, you're probably going to use Altium for everything. This may eventually change but it's going to be a while.

[coppice]

I'm a big fan of KiCad but I think Altium has a bigger lead than you give them credit for. Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon. A large amount of PCB design is not high end, but if you're a company that does 90% simple stuff and 10% stuff that requires a product like Altium, you're probably going to use Altium for everything. This may eventually change but it's going to be a while.

The last time I tried KICAD was a couple of years ago, and it crashed so much I didn't spend much time with it. A little Googling today shows a lot of people highly praising KICAD, but also a number of seasoned people complaining the latest version (5.1) still crashes too much for serious use.

[james_s]

The last time I tried KICAD was a couple of years ago, and it crashed so much I didn't spend much time with it. A little Googling today shows a lot of people highly praising KICAD, but also a number of seasoned people complaining the latest version (5.1) still crashes too much for serious use.

I've been using KiCad pretty heavily for at least 10 years, I can't recall ever having it crash, not even once. I often have 2 or 3 back burner designs I'm working on sitting open in the background for months even. I wonder what people are doing that's causing it to crash? Maybe the Linux version has issues? I've mostly run it on Windows. It has some warts, but stability is not one of them, at least not in my experience.

[apurvdate]

I've seen quite a few big organisations dump Allegro in the last few years, and use Altium across their entire organisation. I'm not sure of the reasons, but Altium is being used for all kinds of exotic designs.

Some Cadence partners are providing the orcad standard license at very aggressive (& attractive) cost throughout last year. Many small organizations who do not work in the fancy stuff are picking up that. They see it as easier path to future scale-up. Even small freelancers are thinking of getting it. A perpetual dongle based license instead of all the cloud & online stuff is great ! This is just another thing eating into altium's share.

[Karel]

Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon.

Neither in altium.

[KE5FX]

I'm a big fan of KiCad but I think Altium has a bigger lead than you give them credit for. Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon. A large amount of PCB design is not high end, but if you're a company that does 90% simple stuff and 10% stuff that requires a product like Altium, you're probably going to use Altium for everything. This may eventually change but it's going to be a while.

Meanwhile...

[daqq]

FFS not another one down the Autodesk drain. Please, anyone, buy them lest they fall under Autodesk turd stained fingers and they give it the Eagle treatment.

[rsjsouza]

I've had better luck with KICAD stability than Altium.

Same here. Not only intrinsic code bugs/failures, but Altium has many more points of failure with network connection for the Vault, licensing, etc. when compared to Kicad.

I recall when our software had paid licensing - the amount of questions and issues from both local and network licenses (the annoying FlexLM software) was immense.

[coppice]

I've had better luck with KICAD stability than Altium.

Same here. Not only intrinsic code bugs/failures, but Altium has many more points of failure with network connection for the Vault, licensing, etc. when compared to Kicad.

I recall when our software had paid licensing - the amount of questions and issues from both local and network licenses (the annoying FlexLM software) was immense.

The licencing with Altium can be a real pain when you are travelling. One of the biggest benefits of free software for people on the move in big multi-nationals is being able to run software whenever you like. Things like Matlab are the worst. Their geolocked licences mean your company can have lots of idle licences in Europe, and American, and India, and Japan, while you are in China and can't use it. When I was friends with people in an EDA support team some years ago, a huge amount of their work was sorting out licence issues.

[olkipukki]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

[Karel]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

Do I understand correctly that the standard license artificially limits the usage (geolocking) and for an extra payment they remove this limitation?

Reminds me to DSO's with license codes that enables certain  options...

[rsjsouza]

I've had better luck with KICAD stability than Altium.

Same here. Not only intrinsic code bugs/failures, but Altium has many more points of failure with network connection for the Vault, licensing, etc. when compared to Kicad.

I recall when our software had paid licensing - the amount of questions and issues from both local and network licenses (the annoying FlexLM software) was immense.

The licencing with Altium can be a real pain when you are travelling. One of the biggest benefits of free software for people on the move in big multi-nationals is being able to run software whenever you like. Things like Matlab are the worst. Their geolocked licences mean your company can have lots of idle licences in Europe, and American, and India, and Japan, while you are in China and can't use it. When I was friends with people in an EDA support team some years ago, a huge amount of their work was sorting out licence issues.

We never did enable FlexLM geolocation and we used to have a grace period of (I think) 30 days before requiring re-connecting. Despite this, the whole scheme can still be prone to failure.

On the other hand, I understand why such things exist and how the software business is a very hard one. Just a handful of companies actually make the big bucks, but the vast majority vanishes quite quickly or become stale. 

[coppice]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

That reduces the pain. However, there's always the trip you take not expecting to need a licence, and then someone asks you to look at something as you travel. If everyone holds a licence every time they travel they will quickly use up all the licences in most companies.

[SiliconWizard]

Where did you hear that?
Protel for DOS was written from scratch by Nick Martin in Borland Pascal, and then moved to Delphi once they moved to Windows. It never used Fortran.

Yep, I can't see why there would be any Fortran in that given the history and platform.

Do you know if AD is still all written in Delphi?

[langwadt]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

Do I understand correctly that the standard license artificially limits the usage (geolocking) and for an extra payment they remove this limitation?

Reminds me to DSO's with license codes that enables certain  options...

in both cases it means those who need those options get to pay for them, those who don't, don't

almost all commercial software is "artificially limited", anyone can download the install files but it won't run without a license


 

[Karel]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

Do I understand correctly that the standard license artificially limits the usage (geolocking) and for an extra payment they remove this limitation?

Reminds me to DSO's with license codes that enables certain  options...

in both cases it means those who need those options get to pay for them, those who don't, don't

No, it means that altium has to pay programmers to write more code (for geolocking) for customers who pay less.
It's the world upside-down.

[langwadt]

The licencing with Altium can be a real pain when you are travelling.

On Demand license + Roaming option = AD Pain Free Usage 

Do I understand correctly that the standard license artificially limits the usage (geolocking) and for an extra payment they remove this limitation?

Reminds me to DSO's with license codes that enables certain  options...

in both cases it means those who need those options get to pay for them, those who don't, don't

No, it means that altium has to pay programmers to write more code (for geolocking) for customers who pay less.
It's the world upside-down.

it means those that only need it in one location gets it cheaper than those that need multiple locations

[coppice]

it means those that only need it in one location gets it cheaper than those that need multiple locations

I'm only ever in one location when I'm using a software package.

[james_s]

Reminds me to DSO's with license codes that enables certain  options...

I like that if it means I can get the scope cheaper by skipping on optional features that I don't need.

I like it even better if I can hack it to enable those features and feel like I got something for nothing even if those features are useless to me, without costing the company anything more to provide it.

[nctnico]

it means those that only need it in one location gets it cheaper than those that need multiple locations

I'm only ever in one location when I'm using a software package.

But you'll still need to replace your PC at one point. I bought a USB dongle with the CAD package (not Altium) I'm using. In the end you don't know how long Altium's servers will be kept up & running to support old licenses which are not under a maintenance contract. Especially when taken over by a company that pushes the subscription model agressively. OTOH you can likely find a cracked version so at least you can keep using the software you paid for.

[langwadt]

it means those that only need it in one location gets it cheaper than those that need multiple locations

I'm only ever in one location when I'm using a software package.

so you pay less than those who need multiple locations

[EEVblog]

Do you know if AD is still all written in Delphi?

Certainly not all of it. I've heard a few years back that they had changed the majority over to C++ or whatever it is.
Given that is was supposed to be 15 million lines of Delphi code in 2014, I'm sure there is still some left:
https://blogs.embarcadero.com/altium-designer-15-000-000-codelines/

As of 2018 it was still using a mix of Delphi, C++ and C# accoridng to the SDK:
https://www.altium.com/documentation/altium-dxp-developer/an-overview-of-the-altium-sdk

I wonder if they still use Morfik for web services? 
Given that the Altium CEO wrote Morfik though...

[T3sl4co1l]

But you'll still need to replace your PC at one point. I bought a USB dongle with the CAD package (not Altium) I'm using. In the end you don't know how long Altium's servers will be kept up & running to support old licenses which are not under a maintenance contract. Especially when taken over by a company that pushes the subscription model agressively. OTOH you can likely find a cracked version so at least you can keep using the software you paid for.

Altium licenses don't run out.  Single user?  Save a local copy and use it forever, only the subscription service runs out (updated vault stuff, software updates, support).

AD also isn't geolocked, at least as far as I know, and not in the way people seem to accuse it of being.  It's up to the customer how they wish to distribute their licenses, and if that ends up having a geo-locking effect, that's essentially their fault.

The sharing mechanism is a license server, operating over the LAN.  If you need to travel, then either you need to VPN into it (or use some equivalent tunnel to access the service), or you need to reserve that license for the duration (roam), in which case it will be unavailable to others whether you're using it at that moment or not.  It's only in this sense it might be considered "geolocked".

AFAIK, it is not disabled by not having internet access, or by travelling internationally, or by connecting to a different network, or by using it on a different machine, etc.

Even if multiple users are detected using the same license, just a nag dialog pops up every 10 min or so, it doesn't lock out either user.

You could just as well "steal" by firewalling AD's internet access and using local license copies on each machine, of course you can't access online services that way (vaults/updates/etc.).

Mind, this is my experience as a US user, I don't know if they have other terms and restrictions in other markets, or other types of licenses besides the standard "seat".  YMMV.

Tim

[Batang]

Promo

[Gerhard_dk4xp]

I'm a big fan of KiCad but I think Altium has a bigger lead than you give them credit for. Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon. A large amount of PCB design is not high end, but if you're a company that does 90% simple stuff and 10% stuff that requires a product like Altium, you're probably going to use Altium for everything. This may eventually change but it's going to be a while.

Meanwhile...

done with Orcad STD on  DOS / Compaq286 @ 16 MHz     
 
Gerhard

[coppice]

I'm a big fan of KiCad but I think Altium has a bigger lead than you give them credit for. Nobody is going to be designing mobile phones, tablets, laptop motherboards or other high complexity stuff in KiCad any time soon. A large amount of PCB design is not high end, but if you're a company that does 90% simple stuff and 10% stuff that requires a product like Altium, you're probably going to use Altium for everything. This may eventually change but it's going to be a while.

done with Orcad STD on  DOS / Compaq286 @ 16 MHz     

Gerhard

I feel a Yorkshireman sketch coming on....We used to lay out dense 24 layer boards with black tape and mylar sheets, and we used to push those tapes 5 miles, up hill, both ways. Thank goodness those days have gone. Boards were a nightmare to get right, and every time we needed to modify a design, the drawing office people nearly cried.

Its interesting that Kicad is being used for quite complex boards, but it says nothing about whether this is an efficient, cost effect, way to work in 2021.

[Karel]

Its interesting that Kicad is being used for quite complex boards, but it says nothing about whether this an efficient, cost effect, way to work in 2021.

Its interesting that Altium is being used for quite complex boards, but it says nothing about whether this an efficient, cost effect, way to work in 2021.

[james_s]

What *is* typically used for high end boards? The last company I worked for that made hardware used Altium IIRC, they were making settop boxes which were essentially computers. Back in the early 2000's I worked at a place that was making another device that was essentially a custom PC motherboard, they were using OrCad.

[Karel]

Zuken, Cadence, ADS

[asmi]

What *is* typically used for high end boards?

Cadence Allegro. Since it costs a fortune, I doubt many people use it to design STM32 boards or Arduinos.
Also pretty much all middle-to-high-end SoCs I came across have their reference designs done in Allegro. I don't think that's a coincidence.

[free_electron]

Where did you hear that?
Protel for DOS was written from scratch by Nick Martin in Borland Pascal, and then moved to Delphi once they moved to Windows. It never used Fortran.

Yep, I can't see why there would be any Fortran in that given the history and platform.

Do you know if AD is still all written in Delphi?

c#. al the old codebase is gone.

[free_electron]

What *is* typically used for high end boards?

Cadence Allegro. Since it costs a fortune, I doubt many people use it to design STM32 boards or Arduinos.
Also pretty much all middle-to-high-end SoCs I came across have their reference designs done in Allegro. I don't think that's a coincidence.

The reason behind that is that the silicon is designed in Cadence tools (which are different from the PCB tools) . The silicon market is owned by Cadence and Mentor.
The silicon tools can export data to the pcb tools. Things like drive strengths, node loads, internal wire lengths for x-net matching.
If you are a semiconductor maker you pay millions in license fees a year for the semiconductor tools and you get the pcb tools for free.

so there's two factors :
1) the tool is free for those people
2) the data exchange for all the stuff like lpddr4 and other critical stuff where information from the silicon is needed.

That is the ONLY reason it is being used. If you just need the pcb tools the license fees are astronomical. The learning curve is also a huge roadblock. They cling to archaic concepts like padstacks. The internal database format is not backward compatible and many more issues.
They dont' want regular pcb-only users. They borged OrCad to get their hands on the masstek codebase , ported it and then sent a 'dear john' letter to all the users that the PCB was going to be discontinued and you needed to switch to Allegro and Concept HDL.

The methodoly used is so different from what we are used to in the Windows world. It is clearly noticeable they have their pedgree in complex unix based tools with extensive customisation using arachic languages like Skill and TCL. The menu system is complex and inefficient and the whole UI totally sucks. It reeks of 1980's Solaris UI design.

[coppice]

If you are a semiconductor maker you pay millions in license fees a year for the semiconductor tools and you get the pcb tools for free.

If that were true, there wouldn't have been a move in the early 2010s by some major silicon vendors to use Altium for all their board work.

[asmi]

That is the ONLY reason it is being used. If you just need the pcb tools the license fees are astronomical. The learning curve is also a huge roadblock. They cling to archaic concepts like padstacks. The internal database format is not backward compatible and many more issues.

The learning curve isn't that bad. Orcad/Allegro has one huuuge advantage over Altium - it does. not. crash. In all my time using it I've never came across any issues or bugs, but in my time with Altium I stumbled over quite a few, not to mention how many times did it crash. Also Altium's SI tools is utter garbage compared to what's included even in 5k$ Orcad Professional, while Allegro is simply in another league. SI in AD is so bad that I have to export my designs from AD into Orcad so that I can do SI sims. I was told my Altium that they are going to improve it in the version 22, so we will see.

They dont' want regular pcb-only users. They borged OrCad to get their hands on the masstek codebase , ported it and then sent a 'dear john' letter to all the users that the PCB was going to be discontinued and you needed to switch to Allegro and Concept HDL.

Yes they do want pcb-only users. Which is why they got rid of old Orcad and just use Allegro engine across entire lineup. That's something Altium should learn from them.

The methodoly used is so different from what we are used to in the Windows world. It is clearly noticeable they have their pedgree in complex unix based tools with extensive customisation using arachic languages like Skill and TCL. The menu system is complex and inefficient and the whole UI totally sucks. It reeks of 1980's Solaris UI design.

And yet somehow it's PCB editor is FAR superior to what's in Altium - it's faster, more efficient and RELIABLE.

[apurvdate]

They dont' want regular pcb-only users. They borged OrCad to get their hands on the masstek codebase , ported it and then sent a 'dear john' letter to all the users that the PCB was going to be discontinued and you needed to switch to Allegro and Concept HDL.

That in fact was better move IMHO. The Allegro interface is way better than old OrCad Layout.
Also I don't need much to learn extra stuff related to ui menus or how a tool interface is, even if I scale up my tool from OrCad Standard to OrCad Professional or even if I move from basic OrCad to higher package Allegro.

[nctnico]

They dont' want regular pcb-only users. They borged OrCad to get their hands on the masstek codebase , ported it and then sent a 'dear john' letter to all the users that the PCB was going to be discontinued and you needed to switch to Allegro and Concept HDL.

That in fact was better move IMHO. The Allegro interface is way better than old OrCad Layout.

I'm not sure I would call the interface better perse but it is hard to compare since the UI design philosophy between Orcad Layout and Allegro are so far apart. But Allegro is a whole lot faster compared to Orcad Layout. Designs that Orcad Layout is struggling with (resulting in slow redraws) are a piece of cake for Allegro. For what it can do the minimum system requirements of Allegro are really low. I can easely work on a complex SOC design on a core i3 laptop which is nearly a decade old.

[Karel]

The methodoly used is so different from what we are used to in the Windows world. It is clearly noticeable they have their pedgree in complex unix based tools with extensive customisation using arachic languages like Skill and TCL. The menu system is complex and inefficient and the whole UI totally sucks. It reeks of 1980's Solaris UI design.

Sounds interesting. Looks like I'd like it. Everything that doesn't smell/look like windows is a huge advantage already.

[rsjsouza]

The methodoly used is so different from what we are used to in the Windows world. It is clearly noticeable they have their pedgree in complex unix based tools with extensive customisation using arachic languages like Skill and TCL. The menu system is complex and inefficient and the whole UI totally sucks. It reeks of 1980's Solaris UI design.

Sounds interesting. Looks like I'd like it. Everything that doesn't smell/look like windows is a huge advantage already.

It may be nowadays, but in the turn of the century I used Mentor tools in a similar capacity (Si and some PCB design) in Sun Ultra 1 stations running Solaris. It was really painful and the UI was very broken (those typical rolling bars in ill-fitted contents in those dialog and options boxes, for example). That gave me an entire new appreciation for the polish of the applications in the then current Windows 98 and 2000. This was one of the reasons why we called it "torMentor".

Fast forward fifteen years, I started dabbing in Cadence at work and the whole interface is not as bad  but it still has that similar feeling. I guess that past experience kinda ruined for me. More or less at the same time I started using Altium in a limited capacity and find its interface better. Performance wise, though, the use of Vault in a networked environment is very upsetting.

[free_electron]

i've seen multiple tools evolve over many years and used them over that timespan.
Smartwork -> Hiwire (dos)
Orcad 3.xx for DOS to Orcad 11 for Windows (dos to win)
Cadence Allegro (sun/win)
Autotrax 1.61 to Altium Nexus Server. (dos to win)
as well as a number of esoteric systems
Valid (sun)
Visula (sun)
Bentley Systems (sun)
EDS (sun)
Teradyne Vanguard (dos)
Intergraph (Clipper)

They all had their pro's and cons.
I do not like the amount of work involved to create a footprint in Allegro, nor it's ui. it still feels like old Sun software
In my opinion you simply can't beat Atlium in terms of productivity. You can whip up a design very quickly. If you combine the tool with a vault installation to do concurrent design, library management and version control, you boost productivity even more.
Link it to your PLM system, suck in datasheets through octopart, exchange live 3d data with the mechanical world. (cad connector to solidworks and others ) and you crank it up another notch.

Does it do signal integrity and power integrity ? No, but link it with dedicated tools for that and you are set. Does it do simulation ? yes ( again) . I'd rather they focus on shcematic/pcb than waste resources chasing an SI /PI simulator. Leave that to the dedicated tools.

Does it crash ? rarely. i'm running this thing 24/7 leaving the session open. I rarely have to restart it (keep an eye on the bottom right resource monitor. if it goes red , either save all and a quick Altium restart, or close all the other gunk you have open. This is a windows issue and not always an altium issue. We are in 2021 and windows is still stuck to 64K UI handles for all processes combined.. The garbage collection doesn't always work properly on those either. The latest versions of Altium have gotten much better at avoiding that misery.

The vault has never restarted in the 5 years i've been running it (and there's hundreds of floating licenses with over 400 people using it)

Does altium become unstable ? yes.
When ? From my experience supporting the installation ( like i said hundreds of licenses and 400+ users)

- underpowered hardware
- glitchy hardware like consumer grade or self built , overclocked stuff.
- graphics drivers (ATI / AMD is problematic. Get a Quadro with certified drivers )
- emulation ( on macbooks )

I'm running the latest version on a 6 year old Zbook 17 G2 with a K3100 quadro. Connected over VPN to a Nexus server backend. Whenever i get complaints from users that it is unstable my first question is : what do you run it on ? if it is not workstation grade machinery (zbook or Z8xx series machines) they get my standard "go to IT and request a real computer" answer. Those that do listen , never come back.
5 year old Zbooks can be had on ebay for as low as 400$ ( and that includes a quadro graphics board).

[asmi]

In my experience Altium really doesn't like sleep or hibernation. I don't ever turn off my PC (except a couple of times a year for cleaning, or every once in a while for a component installation/removal), but I let it sleep/hibernate, and whenever it wakes up, there is close to 50/50 chance that Altium is going to be messed up and will need to be restarted. While Orcad PCB editor often stayed open for over a month with sleep and hibernation, and never had any problems. And none of the issues I had with Altium were caused by PC itself. And Windows is obviously not a problem too as Orcad manages to work for a very long time just fine.

[tszaboo]

In my experience Altium really doesn't like sleep or hibernation. I don't ever turn off my PC (except a couple of times a year for cleaning, or every once in a while for a component installation/removal), but I let it sleep/hibernate, and whenever it wakes up, there is close to 50/50 chance that Altium is going to be messed up and will need to be restarted. While Orcad PCB editor often stayed open for over a month with sleep and hibernation, and never had any problems. And none of the issues I had with Altium were caused by PC itself. And Windows is obviously not a problem too as Orcad manages to work for a very long time just fine.

I dont think it is hibernation, I think it is changing video cards and or outputs. Put my desktop into hibernation for months without restart, no issues. Got a laptop with two video cards, constant crashes and errors when doing that.

[PlainName]

I've had issues with more than one monitor where one of them is a Samsung and the screensaver has kicked in - on waking up the Samsung causes Windows to move and resize any open windows. Twice. Maybe it's something like that.

[free_electron]

In my experience Altium really doesn't like sleep or hibernation. I don't ever turn off my PC (except a couple of times a year for cleaning, or every once in a while for a component installation/removal), but I let it sleep/hibernate, and whenever it wakes up, there is close to 50/50 chance that Altium is going to be messed up and will need to be restarted. While Orcad PCB editor often stayed open for over a month with sleep and hibernation, and never had any problems. And none of the issues I had with Altium were caused by PC itself. And Windows is obviously not a problem too as Orcad manages to work for a very long time just fine.

I dont think it is hibernation, I think it is changing video cards and or outputs. Put my desktop into hibernation for months without restart, no issues. Got a laptop with two video cards, constant crashes and errors when doing that.

That is a hardware / driver issue. If one monitor is on the embedded graphics and the other is on the discrete gpu it does indeed give problems. if you hibernate, close the lid ( but keep running external through a docking station ) the system is thrown fro one graphics driver to another. if they don't support the same feature set in terms of hardware acceleration your application can crash ( and not only altium ... )
When the application starts it interrogates what hardware it runs on and decides to use one GPU library. At startup the JIT engine 'compiles' this library in and the application starts. if you now all of a sudden switch graphics hardware your program is using a graphics library that is not made for that hardware ! Ideally this should work if you are using only the basic , common functions of a graphics card. But once you use the accelerators .. are bets are off. Here run this heavily optimized code for an ATI card (using ATI specific libraries) on an nvidia. it doesn't work. You cannot expect an application to dynamically switch hardware configurations. It is an unreasonable scenario. if it is removable hardware like usb to ethernet dongles etc the OS has a mechanism. But a graphics card is not a 'removable' device.

If you run altium on such a system and drag the window from one screen to another ( or set it in the middle between two , you will only be useing the common windows graphics library. no acceleration will be started. if you start altium on the real GPU , acceleration launches. drag it to the other window and acceleration ceases for the remainder of the session. do this from the OS side (hibernation ,turnign the graphics card off ) without signalling the application : all bets are off.

That's why ,on the Zbook i use the discrete graphics is turned off. it is disabled in the bios. only the Quadro shows up. Two monitors on the dedicated displayports (doscking station) . One on the Thunderbolt port that switches to mini displayport mode.
I've seen all these issues that are mentioned and, like i said before, my standard answer is : get a real computer. in the sense : get a workstation grade machine with fixed hardware and none of this 'dynamically can change' or 'emulated' ( wine, parallels , bootcamp ) environments. You are running CAD software, not video games. I had engineers mucking around with altium on a 13 inch macbook... complaining it is slow. well duh -facepalm- . besides you have to be insane running it on 13 inch screens. back in the dos days the standard was already 14 inch.

What's next ? altium for Iphone ? If we are going that route i also want it on flipphones with 320x240 like motorola razor (it has a second 64x64 display. dual screen ! schematic one , pcb other)?  (no , sorry we will not support blackberry , don't ask. Eagle already has that market  )
how about nokia 3310 ? that certainly needs to be supported ! such an ubiquitous platform !
Maybe we should run it on HD44780 'graphics cards'. That a known standard that's been around for ages. Lets go for 2x8 char configuration. they are cheap on ebay . less than a dollar. let's hook up a few hundred to an arduino and see if we can make it run altium. We already did Doom, so how hard can it be ?

[Bud]

I ran Altium for 14 months on a Win7 PC with Nvidia extension GPU card without rebooting or restarting, just hibernation. That was Altium 17 though.

[asmi]

I've seen all these issues that are mentioned and, like i said before, my standard answer is : get a real computer. in the sense : get a workstation grade machine with fixed hardware and none of this 'dynamically can change' or 'emulated' ( wine, parallels , bootcamp ) environments. You are running CAD software, not video games. I had engineers mucking around with altium on a 13 inch macbook... complaining it is slow. well duh -facepalm- . besides you have to be insane running it on 13 inch screens. back in the dos days the standard was already 14 inch.

How come every other software manages to work just fine on my system (including Orcad PCB, and even - God Forbid - KiCAD!), and the only one which has problems is Altium? It tells me that the problem is not a PC.

What's next ? altium for Iphone ? If we are going that route i also want it on flipphones with 320x240 like motorola razor (it has a second 64x64 display. dual screen ! schematic one , pcb other)?  (no , sorry we will not support blackberry , don't ask. Eagle already has that market  )
how about nokia 3310 ? that certainly needs to be supported ! such an ubiquitous platform !
Maybe we should run it on HD44780 'graphics cards'. That a known standard that's been around for ages. Lets go for 2x8 char configuration. they are cheap on ebay . less than a dollar. let's hook up a few hundred to an arduino and see if we can make it run altium. We already did Doom, so how hard can it be ?

OMG your excuses are getting more and more pathetic Altium has some good sides, but stability is definitely not one of them. And no, computer has nothing to do with that.

[free_electron]

How come every other software manages to work just fine on my system (including Orcad PCB, and even - God Forbid - KiCAD!), and the only one which has problems is Altium? It tells me that the problem is not a PC.

They are not using Acceleration on the GPU. Simple as that. Does OrCad or Kicad support multiple windows served form one application ? split between screens ? Orcad doesn't. Kicad most likely doesn't either. Altium requires hardware that is directx10 capable. what if the directx drivers or version is different for the two graphics cards you have , and the machine switches from one to the other while going into hibernation. you can't expect the software to handle such situations. Who runs CAD on a laptop anyway ? that's what you have workstation grade machinery or desktops for.

I support a pool of over 500 users and 200+ licenses.
if there is trouble it is invariably with people running it on macbooks , office machines and whatnot. Give these people workstation grade hardware (Z8xx series and Zbook)  and there are all of a sudden no more complaints. There may be other machines that work too , but the Z family is readily available for engineering pool due to them being certified for Solidworks and Catia.

I leave Altium running all the time on my Zbook.. it's been up for months without a single hiccup. i can come and go as i wish. The last time i shut it down was to do an upgrade to a newer version. Not gone down since. And Altium work is all i do day in and out. I have mulitple projects open at a time. some of these are 20+ layer boards with more than 15000 design rules , 50000+nets and 10000+ parts , blind and buried vias ,stacked lasered vias. It just works.
That is my experience.
And yes, is see those crashes and weird things too, like people report on the forums. The users i support experience that too , but,  like i said, the moment i switch them to a workstation grade machine with a real GPU ( and the intel disabled) or to a desktop the problems disappear. So explain me how it is not hardware/driver related ? change the machine and the problem is solved. Same operating system.

[Monkeh]

They are not using Acceleration on the GPU. Simple as that.

But they do.

Altium requires hardware that is directx10 capable.

And now we see part of the problem..

moment i switch them to a workstation grade machine with a real GPU ( and the intel disabled) or to a desktop the problems disappear. So explain me how it is not hardware/driver related ? change the machine and the problem is solved. Same operating system.

Do you complain that wet roads are the problem because you have bald tyres?

Hardware varies. Driver versions, vary. Most software manages to work despite this, because they put some effort into error handling and checking for edge cases, and use libraries which supply proper abstraction. Altium doesn't need to be properly engineered because, what, it's engineering software?

Those 'workstation class' machines are exactly the same important silicon as everything else. Running the same shoddy drivers from the likes of nVidia.

[nctnico]

They are not using Acceleration on the GPU. Simple as that.

But they do.

Altium requires hardware that is directx10 capable.

And now we see part of the problem..

Probably. DirectX is for gaming. Orcad etc use OpenGL directly. The problems with Altium are likely shoddy programming and not adhering fully to the specs (if these specs are even fully published by Microsoft). Every now and then I look over the shoulder of someone using Altium but crashes and long waits are frequent and he does use a decent machine.

[Gerhard_dk4xp]

Quote from: free_electron on Today at 01:59:46 am

    They are not using Acceleration on the GPU. Simple as that.


Sometimes I don't care about acceleration. I cannot even see a board
in 2D-mode without directX, something I could do on the Compaq-286 with
the board in post #81 in a previous life.

And my tiny Dell XPS13 has at least 100 times the computation power
of that ancient Compaq and it has a 2000 pixels wide display that plays
HD movies without the slightest slowdown, but AD cannot even do a 2D
display of a postage stamp sized board? And it does not say "directX missing",
no, it says nothing and does simply nothing.

I do not propose to work on an XPS13, but it is handy if you are in the
Icelandic dessert with the motorbike and a customer has some questions.
The alternative is never to be in the Icelandic dessert with the motorbike.
And the XPS13 runs everything else in the VMware machine.

The impression that AD runs more stable on large machines comes
from the fact that they come with large RAM and if there are some GB
to throw away it takes longer until the memory leaks show consequences.

And even on the workstation it sometimes comes to a grinding halt,
like when doing an eco and fetching a new part to the PCB from
a local library takes a minute. For each part.
Closing an unrelated project then CAN do wonders.

And all those nilpointer reads/assignments! I have posted some
of them on the Altium website. These precious little hints are
complained at by the compiler runtime, not the program itself.

Checking for error conditions is very un-altiumish. They obey
strictly the rule "Never check for errors you do not know to handle!".

I remember that  AD16 would suddenly crash when starting it.
I was denied any help with the sorry excuse that I ran it in a Win7
machine under VMware Workstation. They insisted that VMware
workstation could not handle network connections correctly.
Notwithstanding that there was no network involved.

The 1Meg$$$ question: what is the core business of VMware?

I had to find it out myself. It turned out that I had renamed
an old decal library from .lib to  .pcblib  and AD did not
even check the library format - and preferred bombing.
I would not bet that the error is not happily preserved in AD21.

I then stopped my subscription and when I gave in in summer 2020,
everything was different, nothing was better, with new bugs.

For example, it cannot import Gerber files from Keysight Advanced
Design System correctly, it just creates $SOMETHING that has not
the slightest thing in common with the carefully electromagnetics-
optimized original filter design. Import via DXF gives at least a
skeleton that can be filled by hand. Hand coloring of pictures
with numbered points is not my idea of design automation.

It also creates funny nets between unrelated metal segments that can
be simply ignored when running DRC.  It's enough to remember that
14 unrouted connections are OK.
Gives me confidence when I need to import something complicated.
The free and open source gerbv can import Gerbers from ADS
with no issues.
 
Is there any true electromagnetics simulator that can exchange
data with AD?

What has happened to FPGA support? It started slow and then seems
to have starved.

In the spice realm, LTspice is still king.

I don't have the slightest interest of storing my designs in
somebody's cloud. The data is mine, mine only and the customer's.

They want another € 2100 from me next week  for another year of
support that has never ever solved a problem for me since
the Protel days. Guess what I won't do?

Gerhard

[asmi]

They are not using Acceleration on the GPU. Simple as that. Does OrCad or Kicad support multiple windows served form one application ? split between screens ? Orcad doesn't.

Now you are displaying more ignorance  Of course it does. And, unlike Alitum, Orcad does it under both Windows and Linux.

I support a pool of over 500 users and 200+ licenses.

That explains why are you so persistent in defending it, despite your arguments getting more and more ridiculous and pathetic Trust me, you would have way less headaches if you had to support the same amount of users, but using Orcad Pro instead. Because it runs on everything and never crashes. I was able to browse a >600 MB board file of the open handware dual-socket server motherboard on my 7 years old system just fine, let alone on my recently upgraded system (5950X, 128GB of RAM). BTW, it's here in case someone is curious: https://www.opencompute.org/wiki/Server/ProjectOlympus AD would likely choke on a board half that size

[Karel]

And, unlike Alitum, Orcad does it under both Windows and Linux.

Are you using a vm or wine? I thought that Orcad was windows only:

https://www.ema-eda.com/sites/ema/files/resources/files/Cadence%20Allegro%20and%20OrCAD%20Hardware%20and%20Software%20Requirements%20-%2017.2-2016%20Release.pdf

[free_electron]

Trust me, you would have way less headaches if you had to support the same amount of users, but using Orcad Pro instead.

And how many user do you support ? How large is your sample size ? You need to compare like for like. Let's go find someone that has a similar pool and than we can compare. All i hear is lone-users here. Second, people only come to forums when they have issues or to complain. Never to say something good, or when they have no problems. So queue the endless vitriol and hostility towards users that don't run into any problems. It's the same everywhere. I like this hairdryer. Nah that's shit cause my sisters brother in laws mothers aunts daughter burnt her finger on one. I like white bread. No that's bad you should be eating brown bread. I like Ms paint , no you should be using the gimp. I like swimming, i nearly drowned once, walking is much better.

I already told you i see Altium crashes and weird behavior. There is no doubt it has quirks in it and i will never deny that.
But, in my experience, with my user pool, if i switch the user to different hardware the 'crash' issues disappear. Does it have other bugs ? yes , just like any other software. But the dead-stop hard crash disappears with hardware changes.

AD would likely choke on a board half that size

Board size means nothing. Half that pcb is empty space.

[asmi]

Are you using a vm or wine? I thought that Orcad was windows only:

It runs natively on Linux. The guide you linked is specifically for Windows version.

[asmi]

But, in my experience, with my user pool, if i switch the user to different hardware the 'crash' issues disappear. Does it have other bugs ? yes , just like any other software. But the dead-stop hard crash disappears with hardware changes.

LOL. The problem is Altium, not hardware.

[Monkeh]

So queue the endless vitriol and hostility towards users that don't run into any problems.

Like the endless vitriol and hostility you're presenting towards those who don't need to put a different label on the same IC and call it 'workstation grade' to get a machine which works absolutely fine.

I like how you describe bad programming as 'quirks'.

[asmi]

Board size means nothing. Half that pcb is empty space.

LOL. Just 24 DIMM slots is more than I ever seen in any Altium board, nevermind anything else that's on there. Schematic is 230+ pages long.

[free_electron]

Board size means nothing. Half that pcb is empty space.

more than I ever seen in any Altium board

again , how large is your sample size ? There are many many designs you will never get to see that far outstrip that server board.

You should see the development board for the RED 8K camera... they had it at the Altium users conference 2 years ago.

[asmi]

You should see the development board for the RED 8K camera... they had it at the Altium users conference 2 years ago.

That board is nowhere near close to a regular PC motherboard, let alone server boards.

[free_electron]

You should see the development board for the RED 8K camera... they had it at the Altium users conference 2 years ago.

That board is nowhere near close to a regular PC motherboard, let alone server boards.

oh, you were at altium live then ?

[PlainName]

nowhere near close to a regular PC motherboard, let alone server boards

JOOI, is there a range of board.. complexity or whatever? Here you imply a regular PC motherboard is simpler than a server-class motherboard, so I am wondering if there is a workstation-class motherboard which would be, presumably, more complex than a 'regular' PC but less so than a server. An earlier response suggests not, so it would be instructive to know the reality.

[nctnico]

nowhere near close to a regular PC motherboard, let alone server boards

JOOI, is there a range of board.. complexity or whatever? Here you imply a regular PC motherboard is simpler than a server-class motherboard, so I am wondering if there is a workstation-class motherboard which would be, presumably, more complex than a 'regular' PC but less so than a server. An earlier response suggests not, so it would be instructive to know the reality.

All in all board size / compexity doesn't matter. I've seen someone create a reasonably complex SOC board (including DDR memory) using Geda + PCB. Don't ask how much time it took. In the end efficiency of the tool matters. As Asmi wrote: the majority of the development boards for SOCs and other high end chips are not made with Altium; that is also my observation. There has to be a good reason for that. I look over the shoulder of someone designing a SOC board using Altium regulary. Crashes and long waits are frequent which take the speed out of the design process (and he is using a decent workstation PC!). I was involved in selecting Altium for the project based on several parameters but seeing Altium choke so often (and needing a complete re-install of Windows to get it going again at some point with no useable solution from Altium's support), I'm starting to doubt whether it was a good purchasing decission.  If Autodesk is going to takeover Altium they might find that may need to invest far more than they account for to keep momentum going. As I wrote before: KiCad is moving up fast, it has already eaten Eagle and Altium is next.

[thm_w]

Crashing when waking from sleep or hibernation could be memory corruption. Might be worth running memtest and narrowing it down to sleep or hibernate or both, if you really want to investigate it.
My system was unstable and memtest would only catch an error after ~4hrs of testing.

I use sleep with altium/solidworks and have never had a crash from waking. I do not use hibernate.

[asmi]

JOOI, is there a range of board.. complexity or whatever? Here you imply a regular PC motherboard is simpler than a server-class motherboard, so I am wondering if there is a workstation-class motherboard which would be, presumably, more complex than a 'regular' PC but less so than a server. An earlier response suggests not, so it would be instructive to know the reality.

In my opinion, the most complicated part of PC motherboard is RAM interface. So the more memory channels your CPU have, the more complex layout is. The reason is that each channel is over 100 traces with very strict rules as to how they need to be routed (length matching, spacing, etc.). All other typical PC interfaces are now serial (SATA, PCI Express, USB, HDMI, DisplayPort), so they are much easier to route because there are less traces, and they don't need to be length-matched to the same standard as RAM traces are.

With that said, there are HEDT (Hi-End DeskTop) CPUs out there which have 8 memory channels (AMD Threadripper PRO), while there are server CPUs with less channels than that. So it's technically possible for workstation MB to be more complex. Also there are workstations which utilize server-grade CPUs (Intel Xeon CPUs can be found in quite a few of them), which further blurs the line.

But then I made a statement about complexity, I assumed a typical mainstream systems with CPUs that have dual-channel memory controller, compared to the open source server system I linked above that has 2 CPU sockets each having 6-channel memory controller.

I suppose what is complex and what is not is somewhat subjective depending on the kind of boards a person deals with on a regular basis. In my case a typical board has an FPGA on it, x16 or x32 DDR3 memory interface and a bunch of peripheral chips (stuff like Ethernet, ADC/DAC, and other application-specific devices), plus power delivery system for all of that and a system controller MCU which orchestrates everything and makes sure all hardware on board works properly. I don't consider such boards particularly complex, but I imagine that many others will consider boards with 300-400 or more components on them quite complex. Even the very simple FPGA board that I designed in KiCAD for beginners (it's linked in my signature) has 169 components on it, and I regularly receive comments saying it's too complex - despite it's being pretty much as barebones as practical.

[asmi]

Crashing when waking from sleep or hibernation could be memory corruption. Might be worth running memtest and narrowing it down to sleep or hibernate or both, if you really want to investigate it.
My system was unstable and memtest would only catch an error after ~4hrs of testing.

I did right after I got the new system, and let it run for almost 2 days, as it's got 128 GB of memory and so testing that much memory takes freaking forever, but memtest came up clean. Also, like I said, no other applications I regularly use have any problems, and among them are Orcad Pro, KiCAD, Vivado, Visual Studio, and a bunch of VMs which I run from time to time, including those ran through WSL2.

I use sleep with altium/solidworks and have never had a crash from waking. I do not use hibernate.

My new PC has an AIO 360 mm liquid cooler for the CPU (AMD Ryzen 9 5950X), so I don't want to keep it running (and using up the pump's resource) when I don't use it - hence the "hybrid sleep" setup. Thanks to the super-fast PCIE 4.0 Samsung 980 PRO NVMe SSD, PC wakes up almost instantly as soon as I press any button on a keyboard.

[olkipukki]

Are you using a vm or wine? I thought that Orcad was windows only:

It runs natively on Linux. The guide you linked is specifically for Windows version.

Are you running OrCAD under Linux?

[PlainName]

AIO 360 mm liquid cooler for the CPU (AMD Ryzen 9 5950X), so I don't want to keep it running (and using up the pump's resource)

That caught my eye - do these wear out quicker than, say, an electric motor? In a similar ballpark to stuff like flash?

[asmi]

That caught my eye - do these wear out quicker than, say, an electric motor? In a similar ballpark to stuff like flash?

I don't think so, however if it leaks, consequences can be far more serious than from the air cooler fan failure. My previous PC has been up and running for almost 8 years (with some component changes like video card and SSDs), and I hope this one will last at least in the same ballpark.

[asmi]

Are you running OrCAD under Linux?

I'm not, but I think nctnico mentioned that he is. I also remember seeing Linux packages along Windows ones in their download section.

[apurvdate]

Are you running OrCAD under Linux?

As per official documentation only Red Hat Enterprise & SUSE Linux Enterprise Server are supported (i.e. dedicated installer package is available for these). Never used those distros as I mostly use Debian based distros. So can't check if its PCB only or Capture & PCB both.

[nctnico]

Are you running OrCAD under Linux?

As per official documentation only Red Hat Enterprise & SUSE Linux Enterprise Server are supported (i.e. dedicated installer package is available for these). Never used those distros as I mostly use Debian based distros. So can't check if its PCB only or Capture & PCB both.

Offtopic: Only the PCB part runs on Linux (at least for version 17.2). And it runs fine on Debian. After all Linux is Linux; the installation comes with it's own version of all needed libraries so the chance you run into compatibility problems using a different Linux distribution is minimal. I'm doing all the PCB layout work with Allegro running on Linux.

[olkipukki]

As per official documentation only Red Hat Enterprise & SUSE Linux Enterprise Server are supported (i.e. dedicated installer package is available for these). Never used those distros as I mostly use Debian based distros. So can't check if its PCB only or Capture & PCB both.

I'm not, but I think nctnico mentioned that he is. I also remember seeing Linux packages along Windows ones in their download section.

OrCAD came from Windows (and DOS?) world and never run under Linux.
Cadence has stopped using 'original' OrCAD Layout and switched to cut down version of Allegro PCB Designer. Allegro packages used to run UNIXs, so obviously it can run on Linux today.

Conclusion: You cannot run OrCAD under Linux, but I'm happy to hear if I'm wrong

[asmi]

OrCAD came from Windows (and DOS?) world and never run under Linux.
Cadence has stopped using 'original' OrCAD Layout and switched to cut down version of Allegro PCB Designer. Allegro packages used to run UNIXs, so obviously it can run on Linux today.

Conclusion: You cannot run OrCAD under Linux, but I'm happy to hear if I'm wrong

Orcad is now just a name for a license level (four levels actually - OrCAD PCB Standard, Professional, and both of those with PSpice). Actual executable is allegro.exe. But it's still officially called OrCAD (and is sold as such), so, yeah, you can consider yourself proven wrong

And I think this is a great move by them, because it makes going up the license ladder a non-issue - you install a new license, you get access to new features. No reinstallation is required, no learning curve (except for new features obviously). This is something Altium should learn from them IMHO, and instead of creating and maintaining a whole bunch of products, just gate off certain features in lower-tier offerings. This way they only have a single code base to maintain, so there should be much less bugs, and it will be much cheaper too.

[PlainName]

just gate off certain features in lower-tier offerings

Perhaps they've seen what we do with scopes that have that kind of tier gating.

[rsjsouza]

just gate off certain features in lower-tier offerings

Perhaps they've seen what we do with scopes that have that kind of tier gating.

Exactly. This always happened with software.

It doesn't mean they will be able to curb piracy in any way, but they are free to try.

[asmi]

Perhaps they've seen what we do with scopes that have that kind of tier gating.

Well if they do it as a business, then what stops them from just pirating the whole thing right now? Legally there is no difference between hacking you license and just faking it entirely.

But I've never heard of any business which uses hacked scopes.

[Gerhard_dk4xp]

I have hacked my Agilent 54846B. It did run Linux with Openoffice.
OK, running is not the right word. It crawled.

Gerhard

[Uky]

The Allegro (OrCAD) PCB Editor can be executed under both Windows and Linux.
(Separate downloads though) The original Cadence Schematic tool is Concept HDL
which also can be executed under Linux.

The Schematic Editor, old OrCAD "Capture" runs under Windows only.

Cadence released a new schematic tool a few years ago "System Capture".

Looks nice but asking for their strategy, if they intend to replace the old
OrCAD capture and Concept HDL tool with the new one has remained unanswered.
Recently, Cadence added a library converter from .OLB so I guess that
something is in the making.

Cadence also advertised a new product "something" but if it is a new
version 17.4+ (?) or a rebundle of a lot of options I cannot say.

[tooki]

That is not the point. The question is: what is the added value of Altium when (not if) KiCad offers similar functionality?

KiCad might appear to offer similar functionaly to Altium if all you ever do is fairly basic to mid level boards. But if you do very complex mid to higher end stuff, Altium is still streets ahead.

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

[coppice]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

LibreOffice is just as good as MS Office. The problem is they both suck, and need huge amounts of work to make them truly fit for purpose.

[rsjsouza]

That is not the point. The question is: what is the added value of Altium when (not if) KiCad offers similar functionality?

KiCad might appear to offer similar functionaly to Altium if all you ever do is fairly basic to mid level boards. But if you do very complex mid to higher end stuff, Altium is still streets ahead.

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

Although I use Gimp quite extensively for personal purposes and did a lot of work with Inkscape in a professional environment, I agree with you. Gimp's interface is much more aligned with a professional package and easier to use, but Inkscape (as well as Blender, another one that I used extensively in the past) is quite dense. Obviously that reflects the price you pay for it.

Libreoffice is quite alright and has become quite popular to have an ecosystem on itself. MSOffice, which I use extensively at work (2019 version), has a completely different interface (the ribbon stuff) and is polished in some places, but Powerpoint can still ruin your day if you get a presentation from other user (fonts, spacing, page sizes, etc can get messy). 

Kicad has some annoyances, has an interface quite different than a commercial package but the massive improvement over just a few years gives me a lot of confidence it will be capable of dealing with more and more complex designs.

[tooki]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

LibreOffice is just as good as MS Office. The problem is they both suck, and need huge amounts of work to make them truly fit for purpose.

No, it really isn’t. Literally my point: people claim it’s the same without understanding that it is not. Both are good enough for basic needs, but LibreOffice lacks a ton of advanced features used professionally, especially in business workflows. It supports automation in ways LibreOffice can’t dream of, and even where LibreOffice “does” support it, it’s not done with the consistency and reliability of MS Office. (An API is useless if it breaks with every other update…)

Remember this: just because it’s good enough for you doesn’t mean it’s good enough for everyone.

[coppice]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

LibreOffice is just as good as MS Office. The problem is they both suck, and need huge amounts of work to make them truly fit for purpose.

No, it really isn’t. Literally my point: people claim it’s the same without understanding that it is not. Both are good enough for basic needs, but LibreOffice lacks a ton of advanced features used professionally, especially in business workflows. It supports automation in ways LibreOffice can’t dream of, and even where LibreOffice “does” support it, it’s not done with the consistency and reliability of MS Office. (An API is useless if it breaks with every other update…)

Remember this: just because it’s good enough for you doesn’t mean it’s good enough for everyone.

MS Office is utter garbage, not fit for purpose. Its not good enough for anything more than trivial uses. Word especially is like a curse on humanity. That's why most serious document producers don't use it. It would be hard for an alternative to be genuinely worse.

[Nominal Animal]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need.

Mighty strong generalizations there, tooki.  Perhaps you might reconsider exactly what you claim about others there?

I must say, I take serious umbrage at your claim that people like I have "no clue" – your emphasis – what professional use of such programs is.

I designed the first set of collectors cards for the Finnish Defence Forces in 1997, using Photoshop and Freehand.  At the time, I was the webmaster there.  Later, when running a company, I did some full stack web development stuff, and preferred Photoshop; and occasionally optimized artwork for use in official mailings etc, for which I typically used Adobe Illustrator.

I do not consider myself a graphic artist; I don't have the visual style needed for that stuff.  I can do cards and icons, which don't need that much "artistry", and things that really interest me.  I've had a couple of my oil pastels on display, but that's it; I could never do commissions or such.  So, not an artist.  I can definitely do the technical job, down to color separation, masks, and even custom rasterization if I need to.  I've even taught a basic course on image processing (using Photoshop and Illustrator), although that was two decades ago.  Most of my Photoshop work was taking proper artwork from a graphics artist, and optimize them for the task at hand (web, letterhead/watermarks, etc.).

All in all, I have a few years of professional Photoshop and Freehand/Illustrator use.  I'm not much of an artist, only technically proficient.

Nowadays, I use GIMP and Inkscape.  I can do everything I did in Illustrator and Photoshop, in GIMP and Inkscape, although the workflow is a bit different.  For color separation and such, you do need additional (open source, possibly customized) tools.

Yes, the free tools lack spit and polish, and don't have the support and documentation behind them that Adobe can provide.  Many common tasks are much more efficient with Photoshop and Illustrator, so anyone using them for their everyday workflow, definitely should consider using them.  But if you only have GIMP and Inkscape, you can still achieve the exact same end results.

(Now, I really do have a bit of an esoteric career background... so, if you suspect I'm embellishing the details, do contact me via PM or email, and I can provide you with sufficient information for you to verify all of the above.)

[olkipukki]

Word especially is like a curse on humanity. That's why most serious document producers don't use it.

What they are using?

[johnboxall]

Word especially is like a curse on humanity. That's why most serious document producers don't use it. It would be hard for an alternative to be genuinely worse.

So what should we be using?

[T3sl4co1l]

I can't speak for coppice, but I know a lot of professionals use Illustrator or the like.

Which, instead of Microsoft you have Adobe.  Not much of an improvement in terms of corporate scope.

A lot of publishers are perfectly happy working with your Word documents.

Some use \$\LaTeX\$, but as far as I know it's still largely confined to academic (particularly CS, math and physics?) users, and a number of piecemeal automated systems, like, I recall reading there's a train system in Germany that generates their time sheets with scripts and LaTeX.  So, like templated HTML, but flowed nicer.  Given the trouble of using certain packages or formatting (wrapfig most notably), this is probably of limited general utility, but can be a good idea for consistent stuff like that.

Tim

[olkipukki]

I can't speak for coppice, but I know a lot of professionals use Illustrator or the like.

Do you mean FrameMaker?

Illustrator is vector graphics tool,  not sure how it can be Word's substitution... 

[coppice]

I can't speak for coppice, but I know a lot of professionals use Illustrator or the like.

Do you mean FrameMaker?

Illustrator is vector graphics tool,  not sure how it can be Word's substitution... 

FrameMaker has been heavily used for at least 3 decades. It keeps the kind of strict discipline between form and content that serious document producers are looking for. These days I see organisations using a variety of tools, like XMetal, that keep their raw documents in some kind of XML form. Interestingly, most of the people in big prganisations who aren't involved in churning out externally facing documents don't even realise that is how their organisation works.

[tooki]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

LibreOffice is just as good as MS Office. The problem is they both suck, and need huge amounts of work to make them truly fit for purpose.

No, it really isn’t. Literally my point: people claim it’s the same without understanding that it is not. Both are good enough for basic needs, but LibreOffice lacks a ton of advanced features used professionally, especially in business workflows. It supports automation in ways LibreOffice can’t dream of, and even where LibreOffice “does” support it, it’s not done with the consistency and reliability of MS Office. (An API is useless if it breaks with every other update…)

Remember this: just because it’s good enough for you doesn’t mean it’s good enough for everyone.

MS Office is utter garbage, not fit for purpose. Its not good enough for anything more than trivial uses. Word especially is like a curse on humanity. That's why most serious document producers don't use it. It would be hard for an alternative to be genuinely worse.

Except that it’s not. IMHO Word is actually the star of the whole Office package. Spending a bit of time to properly understand it (as I have) results in being very, very productive in it.

Remember that Word isn’t a desktop publishing program, and isn’t trying to be: it’s a tool for serious writing and for relatively simple continuous layouts. (That is, for single documents, not for combining documents like one does in a DTP program.) And for those things, it works very well.

Where people get into trouble is when they try to press a tool into service as something it’s not intended to be.

[tooki]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need.

Mighty strong generalizations there, tooki.  Perhaps you might reconsider exactly what you claim about others there?

I must say, I take serious umbrage at your claim that people like I have "no clue" – your emphasis – what professional use of such programs is.

I designed the first set of collectors cards for the Finnish Defence Forces in 1997, using Photoshop and Freehand.  At the time, I was the webmaster there.  Later, when running a company, I did some full stack web development stuff, and preferred Photoshop; and occasionally optimized artwork for use in official mailings etc, for which I typically used Adobe Illustrator.

I do not consider myself a graphic artist; I don't have the visual style needed for that stuff.  I can do cards and icons, which don't need that much "artistry", and things that really interest me.  I've had a couple of my oil pastels on display, but that's it; I could never do commissions or such.  So, not an artist.  I can definitely do the technical job, down to color separation, masks, and even custom rasterization if I need to.  I've even taught a basic course on image processing (using Photoshop and Illustrator), although that was two decades ago.  Most of my Photoshop work was taking proper artwork from a graphics artist, and optimize them for the task at hand (web, letterhead/watermarks, etc.).

All in all, I have a few years of professional Photoshop and Freehand/Illustrator use.  I'm not much of an artist, only technically proficient.

Nowadays, I use GIMP and Inkscape.  I can do everything I did in Illustrator and Photoshop, in GIMP and Inkscape, although the workflow is a bit different.  For color separation and such, you do need additional (open source, possibly customized) tools.

Yes, the free tools lack spit and polish, and don't have the support and documentation behind them that Adobe can provide.  Many common tasks are much more efficient with Photoshop and Illustrator, so anyone using them for their everyday workflow, definitely should consider using them.  But if you only have GIMP and Inkscape, you can still achieve the exact same end results.

(Now, I really do have a bit of an esoteric career background... so, if you suspect I'm embellishing the details, do contact me via PM or email, and I can provide you with sufficient information for you to verify all of the above.)

Oh, no, I absolutely know for a fact how professionals use those tools, which is precisely why I have strong opinions on the matter: like you, I have an eclectic background. I  worked for years doing professional tech support for Macs (most of which were graphic designers in publishing/prepress), and later worked (as a UX designer) at a web agency, working alongside graphic designers working on digital assets. And the fact is, Inkscape and Gimp don’t even come close to meeting those groups’ needs.

Desktop publishing requires robust support for CMYK color and spot colors; neither one does it. (CMYK export is not the same as native support!). Meanwhile, the web graphics people have been shifting over to specialized programs that do a better job at that than Illustrator, Photoshop, Inkscape or Gimp ever did.

Besides, where did you get the idea that I was talking about you? In my original post, I wasn’t referring to anyone in this thread. And your reply itself points out that Gimp and Inkscape can’t do what the Adobe tools do (“color separation and such”).

[tooki]

I can't speak for coppice, but I know a lot of professionals use Illustrator or the like.

Do you mean FrameMaker?

Illustrator is vector graphics tool,  not sure how it can be Word's substitution... 

FrameMaker has been heavily used for at least 3 decades. It keeps the kind of strict discipline between form and content that serious document producers are looking for. These days I see organisations using a variety of tools, like XMetal, that keep their raw documents in some kind of XML form. Interestingly, most of the people in big prganisations who aren't involved in churning out externally facing documents don't even realise that is how their organisation works.

Though I’ve never had the chance to use it, I’ve heard only good things about FrameMaker, at least for long-form documents like old-school user manuals.

As for it and other structured document systems, I think a lot of businesses would benefit by moving to them. But the fact is, even with them, you need everyday correspondence, reports, etc. And Word rules the roost there. It supports workflow features that LibreOffice just doesn’t do well. I’d love for LO to become an equal, but MS Office has a huge head start, and a much more disciplined developer behind it.

[Nominal Animal]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need.

Besides, where did you get the idea that I was talking about you? In my original post, I wasn’t referring to anyone in this thread. And your reply itself points out that Gimp and Inkscape can’t do what the Adobe tools do (“color separation and such”).

1. Not everyone doing professional work needs CMYK color separation, or other features provided by Photoshop and not GIMP.

2. GIMP is not at feature-parity with Photoshop, no.  Inkscape is not at feature-parity with Illustrator either.  (The opposite is equally true for both pairs.)  Saying they don't even begin to cover the features professionals need, is utter bullshit.

3. I often tell those interested in graphics arts on a computer that they can start with the freely available open source tools like GIMP and Inkscape: everything they can accomplish with expensive tools like Photoshop and Illustrator, they can achieve with other tools as well.  I discourage piracy, and point out that artists suffer from that themselves.

4. I've even shown a not very computer literate person how simple it is to implement ones own filtering programs using various programming languages (I used NetPBM PPM format as the example, natively supported in GIMP), explaining the basic idea of chaining separate tools to achieve desired results, and ones own scripts (including extensions in Inkscape) to do something only oneself can imagine doing.  There can be value in modifying ones tools, and although few artists decide to go that route, knowing it is possible is important.  Similarly, it is important to not get stuck on the idea that you need specific tools to achieve a specific result or quality: it is not true at all.

5. I've shown, I hope, that I very definitely have a clue what professional use of these is.

I'm sorry, tooki, but your claim sounds exactly like someone enamored of their tools, or having assimilated the worldview of people fully enamored of their tools, and cannot imagine using any other tools to achieve the same task.  I worked for years with people who had for example very strong opinions wrt. Freehand versus Illustrator, so much so as to yelling in meetings about it, so I'm familiar with the argumentation style.

I do know, however, how compelling those arguments are in the real world workplace.  Not because they're right, but because they're so loud and persistent, and based on deep emotional convictions.

[PlainName]

Can I just butt in to say my partner is in the publishing industry and the bulk of her work is with Word documents and PDFs. Word is a necessity - Libre Office isn't compatible enough - and people use it mostly because everyone else uses it. OTOH, she uses pdfXchange for PDFs because it's better than the Adobe offering and there is no compatibility issue.

Obviously, Word isn't suitable for doing the layout of the Financial Times, for instance, so a big problem with this discussion is that 'professional' seems to cover anything and everything that might involve words. Too big a field to even start generalising, but to insinuate that Word isn't or can't be used professionally is, and I quote, 'cobblers'.

[Nominal Animal]

To provide an opposite viewpoint to my own above: it is rather common to be tied to specific applications, just because the users demand those particular ones.

If I were hiring graphics artists, I would have to provide them with Photoshop, Illustrator, et. al, simply because those are the de facto standard tools for those tasks.

It has nothing to do with what tools are necessary to achieve specific results, and everything to do with what tools are so popular among workers that they become expected, the de facto standard for that job.

[Karel]

So, will autodesk do another (higher) bid for altium?

[PlainName]

Hope not!

If they bought it would it replace Eagle? Even now after they've embedded Eagle, or its functionality, in Fusion? It would be one thing to buy Altium to expand their product line, but they already have PCB design capability. So perhaps the plan would be to buy it so they can bury it and remove some competition. If so, there must be a limit to how much they can throw away just to leave a clear field. Which wouldn't be that clear with Orcad being in the same ballpark.

[tooki]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need.

Besides, where did you get the idea that I was talking about you? In my original post, I wasn’t referring to anyone in this thread. And your reply itself points out that Gimp and Inkscape can’t do what the Adobe tools do (“color separation and such”).

1. Not everyone doing professional work needs CMYK color separation, or other features provided by Photoshop and not GIMP.

2. GIMP is not at feature-parity with Photoshop, no.  Inkscape is not at feature-parity with Illustrator either.  (The opposite is equally true for both pairs.)  Saying they don't even begin to cover the features professionals need, is utter bullshit.

3. I often tell those interested in graphics arts on a computer that they can start with the freely available open source tools like GIMP and Inkscape: everything they can accomplish with expensive tools like Photoshop and Illustrator, they can achieve with other tools as well.  I discourage piracy, and point out that artists suffer from that themselves.

4. I've even shown a not very computer literate person how simple it is to implement ones own filtering programs using various programming languages (I used NetPBM PPM format as the example, natively supported in GIMP), explaining the basic idea of chaining separate tools to achieve desired results, and ones own scripts (including extensions in Inkscape) to do something only oneself can imagine doing.  There can be value in modifying ones tools, and although few artists decide to go that route, knowing it is possible is important.  Similarly, it is important to not get stuck on the idea that you need specific tools to achieve a specific result or quality: it is not true at all.

5. I've shown, I hope, that I very definitely have a clue what professional use of these is.

I'm sorry, tooki, but your claim sounds exactly like someone enamored of their tools, or having assimilated the worldview of people fully enamored of their tools, and cannot imagine using any other tools to achieve the same task.  I worked for years with people who had for example very strong opinions wrt. Freehand versus Illustrator, so much so as to yelling in meetings about it, so I'm familiar with the argumentation style.

I do know, however, how compelling those arguments are in the real world workplace.  Not because they're right, but because they're so loud and persistent, and based on deep emotional convictions.

My original statement was about people claiming the OSS tools are complete replacements. I said they’re not. You said they’re not. Why are you fighting with me? To be clear since you’re not picking it up: I said nothing about PARTIAL replacements.

[floobydust]

So, will autodesk do another (higher) bid for altium?

Altium apparently rejected the $38.50 then a $40 verbal bid and Autodesk scurried away and terminated discussions a month ago.
A $3.9B deal. Likely mostly ADSK stock, which is on a roll, an excellent bubble. Acquisitions were popular in the 1990's and now it's back. Become Tyrannosaurus Rex so big and greedy, everybody snort coke and watch the investors pour in the money. But I thought ADSK "growth" is fuelled by their massive debt, money is cheap right now just borrow it and scoop up companies to give the illusion of growth until the hedge funds jump in.

Autodesk is hell bent on their push to cloud apps for engineering. Recently bought Upchain, Innovyze. It's all about the MBA fad, recurring subscription plan revenue.
Whatever happened to having a decent core product in the first place as a way to profit, instead of spending everything on the way to milk the cow harder?
All their energy is going into forcing engineers to use their big blob software in the cloud. Even Altium fell into that trap. It's fueling open source.

I've used both Photoshop and GIMP. GIMP is the best way to kill someone, the user interface is so consistently terrible (try draw a circle) that I laugh everytime I use it. But it works, no fees or subscription and not locked into Windows or the cloud. It's a tool and I don't mind lesser features for lesser bullshit.

[nigelwright7557]

In the 1980's there were very few PCBCAD dealers around and what there was was asking quite a lot of money for it.

These days the market is flooded and then there are about 20 free packages around !

[EEVblog]

In the 1980's there were very few PCBCAD dealers around and what there was was asking quite a lot of money for it.
These days the market is flooded and then there are about 20 free packages around !

Anyone remember PCBreeze from Anton Kepic?
https://www.linkedin.com/in/anton-kepic-a0b883a/
That was my first PCB package.

[nctnico]

That reminds me of the people who say that LibreOffice is a complete substitute for MS Office, or that GIMP and Inkscape can replace Photoshop and Illustrator: they have no clue what professional use is, and that the open source apps don’t even begin to cover the specialized features professionals need. They’re perfectly adequate for everyday use in many cases, just not true pro use.

This isn’t to say that open source desktop apps can’t be best-in-class, but IMHO they rarely are, not in functionality, and most certainly not in usability. Quite different from the server stuff, where open source has managed to produce truly outstanding software, just to list one example.

LibreOffice is just as good as MS Office. The problem is they both suck, and need huge amounts of work to make them truly fit for purpose.

No, it really isn’t. Literally my point: people claim it’s the same without understanding that it is not. Both are good enough for basic needs, but LibreOffice lacks a ton of advanced features used professionally, especially in business workflows. It supports automation in ways LibreOffice can’t dream of, and even where LibreOffice “does” support it, it’s not done with the consistency and reliability of MS Office. (An API is useless if it breaks with every other update…)

Remember this: just because it’s good enough for you doesn’t mean it’s good enough for everyone.

MS Office is utter garbage, not fit for purpose. Its not good enough for anything more than trivial uses. Word especially is like a curse on humanity. That's why most serious document producers don't use it. It would be hard for an alternative to be genuinely worse.

I agree with this. If I had to do professional text editing then MS Word would not be on the list of packages I'd consider. It is not fit for professional use (especially since the 2010 version which is seriously broken compared to the older versions). Only a true masochist would use MS Word for large documents.

[m98]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier.

This. Although I've never used Eagle and have no desire to change after trying it out several times, I usually like Autodesk software. An EDA-software with AutoCAD UX would be something I'd really like to see eventually.

[nctnico]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier.

This. Although I've never used Eagle and have no desire to change after trying it out several times, I usually like Autodesk software. An EDA-software with AutoCAD UX would be something I'd really like to see eventually.

You mean one where you can also type commands?

[Karel]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier.

This. Although I've never used Eagle and have no desire to change after trying it out several times, I usually like Autodesk software. An EDA-software with AutoCAD UX would be something I'd really like to see eventually.

You mean one where you can also type commands?

Like in Eagle, I use that a lot. Much faster than clicking with the mouse and it helps also to avoid RSI.
It's a mystery to me why not all EDA packges offer this. The mouse in general is a terrible input device...

[nctnico]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier.

This. Although I've never used Eagle and have no desire to change after trying it out several times, I usually like Autodesk software. An EDA-software with AutoCAD UX would be something I'd really like to see eventually.

You mean one where you can also type commands?

Like in Eagle, I use that a lot. Much faster than clicking with the mouse and it helps also to avoid RSI.
It's a mystery to me why not all EDA packges offer this. The mouse in general is a terrible input device...

In that case you should look at Orcad PCB Designer / Allegro . I never use the mouse to position something accurately (pads, lines, etc); I just enter the coordinates.

[olkipukki]

I've used both Photoshop and GIMP. GIMP is the best way to kill someone, the user interface is so consistently terrible (try draw a circle) that I laugh everytime I use it. But it works, no fees or subscription and not locked into Windows or the cloud. It's a tool and I don't mind lesser features for lesser bullshit.

Its not too bad, you can still find alternatives

 https://affinity.serif.com/en-us/

I'm using these products a few years on both Win and Mac after give up on Adobe Creative Suite, paid once and still receiving updates (!!!), although don't mind pay for a next version upgrade

[olkipukki]

I agree with this. If I had to do professional text editing then MS Word would not be on the list of packages I'd consider. It is not fit for professional use (especially since the 2010 version which is seriously broken compared to the older versions). Only a true masochist would use MS Word for large documents.

What you will use?

[voltsandjolts]

Well, this thread got derailed. Technical documentation tools would be a worthy topic on its own.

I've never worked for a company with dedicated technical authors.
I'm not a Word fan, but it's the lowest common denominator for doing technical documentation, which is often a team effort.
I would get some interesting reactions if I suggested \$\LaTeX\$, asciidoctor, oxygenxml, madcap flare etc. to some other folks here.
Word can make a half decent PDF, but fails badly compared to proper single source - multiple output systems.

IIRC the RPi RP2040 datasheet is done in asciidoctor and it looks great.
I wonder what Microchip, ST, AD etc use for their datasheets?

[PlainName]

Are we not slightly confusing things here? Apart from no longer talking Altium or Autodesk, I mean! ISTM there are two aspects to 'professional' wordsmithing: figuring out what you want to say and making it look how you want (hopefully nice). The tools for each need distinct and different features, so you might well have one tool for writing the stuff and another for outputting the result. Someone arguing that X is fantastic (for collating material and organising it) maybe won't agree when someone else insists Y is far, far superior for (layout and stuff).

[olkipukki]

I wonder what Microchip, ST, AD etc use for their datasheets?

It's depend...

You can have a look the related field (aka source application) in PDF metadata/properties.

[Nominal Animal]

TL;DR: I don't know what will happen, but here is a reason why something like Altium takeover by Autodesk makes a lot of long-term commercial sense for Autodesk; and surprisingly, it reveals an interesting facet of what is happening wrt. PCB design/EDA right now.


One important aspect regarding Altium/Autodesk is that due to the appearance of cheap prototyping services (including surface mount component populating and soldering services cheap enough for pure hobbyists), PCB design is bifurcating into cheap+hobbyist and professional aspects.

(As you probably know, I'm more a software/physics person with only theoretical and butterfinger hobbyist knowledge about electronics, but I do have seen this more than once before: in mid-1990s with multimedia CD production ("true" programming vs. authoring software like Macromedia Director), and in the last two decades in various web site authoring tools (initially single-user offline only, then online; now with entire mainentance stacks up to customizable themes, not only for websites, but starting to focus on specific subjects and approaches like EduX and MIT courseware tools).  So, I'm saying this based on extrapolation to similar situations I've observed before, and not as someone who is actually in the midst of the events.  This means this is extrapolation + speculation, and definitely not reliable information you should trust.)

Considering Autodesks product portfolio, they probably also see this, and are looking to add such a "pair" into their lineup, both as a new subfield to expand into, but also as a way to increase the customer base to their main product lines.  (It is easier to sell packaged software deals, than individual software, to customers who are not already users of your software.  It is somewhat a marketing issue: "think of everything you could do with these tools, now that you're considering trying to do some of this!".  Those that are already users of your software, are more a customer retention issue than a marketing/approach issue.)

The free/cheap but limited (note: not just hobbled, but simplified, so that the tool is easier to approach for new users) package as an introduction to the professional quality costly package, is a tested and true approach.  It is not a gold mine or marketing trick, but it does work to increase market share, and user base for the pricy package.

I do suspect that EasyEDA, a purely web-based no-cost PCB design software, is being carefully observed by the established EDA software vendors.

On one hand, its purely web-based implementation likely seems attractive to pointy-haired-bosses who dream of new revenue sources, without realizing that it is actually the one downside in the hobbyist domain.  (If anyone were to replicate EasyEDA functionality with configurable integration to various PCB manufacturers, using a free standalone Linux, Windows, MacOS, Android application, they could capture the hobbyist market.  However, the price is key in that segment, quite opposite to the professional segment.)  It would be very interesting to know what kind of communications they have had with the various EDA vendors (if at all).

As a butterfinger hobbyist, I do not really know what is happening in the professional segment.  I see that Eagle etc. seem to be still considered way more productive environments than e.g. KiCAD, but I do not know the precise reasons.  The fact that the cheap prototyping services can now do modest-precision four-layer boards (say, BGAs with 0.8mm pitch or larger) at the very low price point, means capability limitations (as used on older Eagle versions, I think) won't really work; the free/cheap packages should concentrate on simplicity and easy to approach user interface, and not "professional experience" with limited features.  Yes, the free/cheap should be a bit of a "toy" compared to the professional ones, because that way those who do start to learn the skills, will upgrade to the professional tools.
(Why that really is so, is related to why most Linux developers do not care about the popularity of Linux, only about its capabilities.  It can feel weird, but there is a sound, robust logic behind it.)

If Autodesk or anyone else asked me, I'd say this: "Provide long-term support for a new, free, offline, open-source EDA project, that implements an user interface as simple and as easily approachable as EasyEDA, with user-customizable integration to PCB prototyping services, and on-line parts catalogs.  For those users who outgrow it, ensure a simple upgrade path to a more powerful professional package.  You can use existing cores, but the user interface you'll probably need to revamp.  Make it modular, so that additional revenue can be obtained by task-specific extensions, like heavy-computation (GPGPU) autorouting, via a subscription service.  Use true UX professionals, observing actual professionals using various EDA tools, to make the user interface efficient.  Forums like EEVBlog are a good source for ideas to test.  The UX is what makes a package stand out, since any other proprietary EDA vendor can create their own as an upgrade from the free package; that competition is necessary to maintain the development efforts long term, without stagnating the product, and one day unexpectedly losing market share catastrophically to a competitor as that competitor was not observed in time."

Although software business is only a few decades old, and we're still "learning" how it really works, software almost always dies in one of three ways: slow stagnation as the domain it targets itself dies out, bad design and management diverting the tool from something users need to something that is no longer worth the cost or effort, or being usurped by a better competition because leadership believes that a leading marketing position once gained is forever held even without development efforts.

Unfortunately, in my experience, the middle management in software development houses is too stupid to see or leverage any of the above.  There is a definite opportunity for both open-source and commercial proprietary software development, but to work financially, especially long-term (as opposed to, creating a company for the purpose of one of the bigger software houses purchasing it), it needs shrewd management that understands more than their own small domain; something I fear is unobtainium in the software sector.

[PlainName]

I'd say this: "Provide long-term support for a new, free, offline, open-source EDA project

It's the offline that's the killer. They won't do it because they lose control and potential revenue stream. Anyone sensible would want it because of the unreliability (not in terms of  connection but because the vendor changes things on a whim, you can't not take an 'upgrade' to the user interface, etc) of 'online'.

For those users who outgrow it, ensure a simple upgrade path to a more powerful professional package.

That would seem to be the sensible route to more dosh. However...

You can use existing cores, but the user interface you'll probably need to revamp.

It would need to be forware and backward compatible. If you can't open your existing projects in the other product, or can't figure out how to operate it without 3 weeks practice, you might as well use a different vendor's stuff.

[nctnico]

Use true UX professionals, observing actual professionals using various EDA tools, to make the user interface efficient.  Forums like EEVBlog are a good source for ideas to test.  The UX is what makes a package stand out, since any other proprietary EDA vendor can create their own as an upgrade from the free package;

That won't work because for the occasional PCB layouter an intuitive UI will be the most productive where such a UI will only get in the way of someone doing PCB design all day.

[EEVblog]

Use true UX professionals, observing actual professionals using various EDA tools, to make the user interface efficient.  Forums like EEVBlog are a good source for ideas to test.  The UX is what makes a package stand out, since any other proprietary EDA vendor can create their own as an upgrade from the free package;

That won't work because for the occasional PCB layouter an intuitive UI will be the most productive where such a UI will only get in the way of someone doing PCB design all day.

I can remember doing a KiCAD first impressions video and pointing out all sorts of thing that might suck for daily use, and I got hammered for it. All the people complaining about my comments I'm sure had never worked 8 hours a day every day for 2 months on a single PCB as a full time PCB professional 

[Nominal Animal]

Use true UX professionals, observing actual professionals using various EDA tools, to make the user interface efficient.  Forums like EEVBlog are a good source for ideas to test.  The UX is what makes a package stand out, since any other proprietary EDA vendor can create their own as an upgrade from the free package;

That won't work because for the occasional PCB layouter an intuitive UI will be the most productive where such a UI will only get in the way of someone doing PCB design all day.

No, the free/non-professional package targets occasional/hobbyists.  For the professional tools, you want the interface to be efficient and productive.  The two are orthogonal.

I can remember doing a KiCAD first impressions video and pointing out all sorts of thing that might suck for daily use, and I got hammered for it. All the people complaining about my comments I'm sure had never worked 8 hours a day every day for 2 months on a single PCB as a full time PCB professional 

That's exactly why I said one needs to observe how actual professionals do use their tools, before you can even start making a truly effective interface.
What people say is not what they do; and those with the least know-how are often the most vocal.  Those who know, know how complex the situation is, so when there is no simple answer, they just prefer to stay out of the mess.
Observing the actual process and tasks, when the observee knows what they're doing, is how you capture the useful practical knowledge.

Hobbyists and professionals use tools in a completely different way.  To hobbyists, ease of use and intuitiveness is important; that way the use is more like playing with something.  Professionals need to get shit done with as little conscious effort as possible, so that they can direct their efforts where it matters the most.

The ideas for user interfaces one collects, are just that: tentative suggestions to try, not something you adopt immediately.  They do need to be experimented on in practice, preferably with the exact kind of users the tool is supposed to target, before their value can be ascertained.  Even if the idea is from a professional, there may be unseen consequences or side effects that make it impractical.
Practice beats theory, every time.

Fact is, the split is such that a single EDA package can no longer fully cater to both professionals and hobbyists.  The cheap prototyping services were the key to bring this about.  A hobbyist is not going to pay $100 for a license so they can design a $5 PCB or two; and they're not going to want to spend much time learning a tool for a similar reason.  Plus, hobbies are supposed to be fun.

Yet, as we know, hobbies often end up affecting ones career, and open up new opportunities.
So, to capture a larger share of the professional tool market, you need to provide a path for the hobbyists to upgrade to the professional tools, if and when they have obtained the necessary skill.

In the past, it sufficed to make sure your tool is the one used in educating new engineers.  This still applies to e.g. office software.  At this point and in the future, relying on that to capture the market share may be too late, because the next generation of EEs is likely to know a range of several EDA tools and not just the one they used at school.

[tooki]

I've used both Photoshop and GIMP. GIMP is the best way to kill someone, the user interface is so consistently terrible (try draw a circle) that I laugh everytime I use it. But it works, no fees or subscription and not locked into Windows or the cloud. It's a tool and I don't mind lesser features for lesser bullshit.

Its not too bad, you can still find alternatives

 https://affinity.serif.com/en-us/

I'm using these products a few years on both Win and Mac after give up on Adobe Creative Suite, paid once and still receiving updates (!!!), although don't mind pay for a next version upgrade

Yep. I’ve been using the Affinity products for years now, since my personal needs do not come close to requiring Adobe. They work well for everything I need to do.

For non-professional use there are tons of options. (My original claim was against people claiming that the OSS packages were complete replacements for the big commercial apps, and I maintain that they are not complete, because they lack features that are critical to professional users. Not every pro user needs every one of those pro features, but without all of them, another tool is not a complete replacement.)

Are we not slightly confusing things here? Apart from no longer talking Altium or Autodesk, I mean! ISTM there are two aspects to 'professional' wordsmithing: figuring out what you want to say and making it look how you want (hopefully nice). The tools for each need distinct and different features, so you might well have one tool for writing the stuff and another for outputting the result. Someone arguing that X is fantastic (for collating material and organising it) maybe won't agree when someone else insists Y is far, far superior for (layout and stuff).

This!!! Absolutely correct, and something I touched on earlier. That’s why I say that Word is a great writing tool. I never said it’s a great layout tool. I feel that a lot of people who hate Word are trying to use it for desktop publishing, which isn’t what its focus is.

Heck, in my daily home/school use, I use Word for writing, but usually Apple Pages for precise layouts. (I can’t justify spending $$$ for Adobe, though it would give me more control.)

[tooki]

Use true UX professionals, observing actual professionals using various EDA tools, to make the user interface efficient.  Forums like EEVBlog are a good source for ideas to test.  The UX is what makes a package stand out, since any other proprietary EDA vendor can create their own as an upgrade from the free package;

That won't work because for the occasional PCB layouter an intuitive UI will be the most productive where such a UI will only get in the way of someone doing PCB design all day.

I think intuitiveness and productiveness are orthogonal. Bearing in mind that I used to work as a UX designer, I think it is possible to design interfaces that have pro-level productivity but are intuitive to new or infrequent users. First of all, intuitive ≠ simplistic. Pro apps have inherent complexity and that can’t be magically eliminated. But neither should pro apps have to tolerate obtuse interfaces (which is often very much implicitly claimed when people argue in support of bad pro interfaces).

To me, a great example of an intuitive pro interface was the old Final Cut Pro. (The ones before the newly developed Final Cut Pro X series, which I have no experience with. I’m not saying that X is harder to use, I simply do not know.) FCP’s interface was so well designed that new users could sit down and quickly become productive, but also had true pro functionality rivaling incumbent systems (like Avid) that had a much steeper learning curve. It had pro-level keyboard shortcuts, so that full-time users could be extremely productive with it. But unlike the harder apps, you didn’t have to memorize them right away in order to accomplish even basic tasks.

Another is Adobe InDesign: within a few years of release, it dethroned the incumbent leading desktop publishing program (Quark Xpress) with the one-two punch of being a better product (having a much better user interface and much better support for then-“modern” features and technologies like alpha transparency, advsnced typography, Unicode, etc.) and for treating its customers far better than Quark did. (Oh, the irony…) InDesign was also easier to use than Adobe’s own existing DTP program, PageMaker.


Anyhow, what I’m definitely not saying is that an intuitive interface is a dumbed-down one. I’m also not saying that a pro app should cater to people with no subject knowledge. A pro app also has every right to expect its users to understand established industry jargon. (In contrast, a consumer app might use a more “everyday” description for the same thing, like how a desktop publishing program will say “leading” while a word processor will say “line spacing”.)


The difficulty in creating great application interfaces is that it takes a ton of effort: from in-depth user observation and testing (as others have already said), to investing lots of time in designing and coding a UI (since good interfaces use code to figure many things out on their own, so the user doesn’t have to tell it redundant information, while still offering a way to change it if need be), but also because a truly good interface dictates some of the internal architecture of the program. Many bad UIs are the result of being a GUI bolted onto an application engine that wasn’t designed for that GUI. This remains a huge issue with legacy systems that were originally designed for batch processing or text-mode terminals. The architecture of the app engine for an interactive graphical interface will be fundamentally different from those. And even within natively GUI apps, the internal architecture depends a lot on the interface semantics.

Often times, the resources to make a truly great GUI app simply aren’t there. Or you don’t want to take the risk entailed by a complete re-engineering of an existing program. (A very valid argument in many cases; that’s why software in industries like banking, insurance, and aerospace, for instance, tends to change things very conservatively and slowly.) And existing GUI programs don’t want to change so much as to alienate existing users, even if the new UI is objectively (as proven empirically in user testing) easier and faster to use for both new and experienced users! (As Microsoft learned with Office 2007.) So you end up with an interface that’s good enough, but eludes excellence.

[ajb]

Heh, as a long time AutoCAD user for a couple years I was hoping AutoDesk would come out with an Altium competitor.  AutoCAD still has one of the best ways of taking a sophisticated set of tools and making them incredibly fast and reliable to use IMO, and I would have loved to see a real EDA tool built around a similar UI philosophy.  Those hopes were dashed when they bought Eagle, and I don't blame them for trying to buy Altium.  It would probably fit nicely into their lineup to have the Fusion 360 w/Eagle tier and then the Inventor/Altium Designer tier.

This. Although I've never used Eagle and have no desire to change after trying it out several times, I usually like Autodesk software. An EDA-software with AutoCAD UX would be something I'd really like to see eventually.

You mean one where you can also type commands?

I happen to have been doing a bit of AutoCAD work these last few days which reminded me that I meant to respond to this.

AutoCAD isn't as good as it is just because it has a command line.  For one thing, it has a *REALLY GOOD* command line.  All of the most common commands are aliased to short mnemonics, it has autocomplete, it does a surprisingly good job of correctly interpreting mistyped commands, and for a while now they've put the autocomplete and command action/option prompts right at the cursor so you almost never have to look at the actual command line.  All of this works especially well because instead of having a ton of specialized commands, most commands have very sensible default behaviors but easily accessible variations/options.  Want to draw a circle? Type "C", then pick your center point and radius point.  Or "C", pick the center point, then "D" and enter the diameter.  Or "C", then "2" for two-point (diametric) mode.  Or "C" then "3" for three-point mode.  Or "C" then "T" for tangent/tangent/radius mode. Don't remember what the options for a command are? Just look at the command line window, or the prompt that appears at the cursor and hit the down arrow to see the list.  This is an incredibly fast way to access a ton of functionality without having to remember a whole lot of arbitrary key combination shortcuts.  (For the record, I think Altium's hierarchical single-key method is pretty good too, but AutoCAD's method absolutely gets you more functionality for your effort.  Altium also has annoying inconsistencies or...antiparallelisms? in the shortcuts for core commands between sch and pcb.  Like, placing text is P,T in sch but P,S in pcb, and making an electrical connection is P,W in sch but P,T in pcb.  Those commands are logically related and, IMO, ought to have the same shortcuts, or at least have shortcuts that don't collide with each other.)  For bonus points, any UI action in AutoCAD, including clicking on graphical buttons, shows up in the command line, so it's super easy to learn the commands for something by clicking around in the UI. Altium has some of that, but not nearly as much.

Speaking of speed, AutoCAD is fast.  It's really, REALLY fast, at least when run on decent hardware.  Granted, 2D line drawings aren't generally that computationally heavy, and it CAN take a second to think on some of the more computationally complicated commands, but the user interface itself is almost always super responsive.  Altium will occasionally miss the second key in one of the shortcuts, I have NEVER had that problem using the command line in AutoCAD.  You would have a really difficult time trying to go faster than it can keep up with except in a few scenarios.

It has an INCREDIBLY good snap/tracking system.  It's lightyears better than anything else I've ever seen.  It doesn't just track from whatever snap points are nearby--if you want to track from a snap point, you hover the cursor over it for a second to TELL it you want to track from that point.  You can choose which types of snap points to allow, which angles they track, and whether those angles are absolute or relative to the last segment.  You can change all of those settings right in the middle of a command with no ill effect.  You can also set temporary track points mid-command, so if you want to track from the point where the extensions of two lines would intersect, you just type "TT", hover over the two endpoints to activate their tracking, track to the intersection, click there, and then you can resume your command with the ability to track off of where you just clicked.  I would pay a significant amount of money just to have this one featureset in Altium instead of the absolute fucking garbage that is its snap/alignment system.  I cannot tell you how much I love AutoCAD's snap/tracking, nor how much I hate Altium's shitty half-baked attempt at it.

It has a solid and fast selection/disambiguation method.  Click to select, shift+click to unselect, drag right to select enclosed/left to select touching (how recently did Altium add that last one?), escape to unselect all.  Altium can sort of be configured to replicate that, but it just doesn't work as well because of other issues with selection.  It's often difficult to tell what's going to get selected when you click on something, and trying to deselect one thing often ends up selecting something else.  In AutoCAD, objects always highlight when you hover the cursor over them.  If the object you want isn't the one that highlights, Shift+Space will cycle through the objects available to click on there, so just tap that until the one you want is highlighted and click.  There's no waiting for an object disambiguation window to pop up (which it doesn't always do when it should anyway).

This is a really fundamental thing and I honestly can't believe that it's worth mentioning, but Altium is actually bad at this: In AutoCAD, things go where I want them to.  I have to constantly fight stupid little positioning errors in Altium, partly because of shitty snap/tracking, partly because of UI lag that means it doesn't register a click until after I've moved the cursor, but also just....sometimes shit doesn't end up where I clicked.  Interactive routing is its own horrible beast in this respect.  A lot of these errors are small enough or on sufficiently insensitive geometry that it doesn't REALLY matter, but it's absolutely indicative of fundamental problems not just with the way they've approached the UI, but with the actual math behind the commands.  I have seen enough weird tiny little offsets in coordinates show up in layouts that I'm convinced the application has deep and widespread problems with rounding at every level of the CAD stack.  That is an ABSURD problem for a CAD program to have*.

There's also just a ton of needless friction in the Altium UI.  Just today I was fixing my display modes (because I had to reload my saved preferences because one specific keyboard shortcut just stopped working and that's the only way I could find to fix it), and the transparency matrix requires you to click on a cell in the grid, and then click into the box at the bottom to change its value.  There's no multi select, so you have to do that on every. single. cell. you want to change.  Who the fuck made that UI decision?  Why can't I change the value in the cell itself?  Why can't I change multiple values at once?  This is hardly an isolated issue, there are needless impediments to being productive all throughout the application.

I could really go on and on about this, but the last think I'll say about the comparison:  Altium has a ton of bugs.  AutoCAD doesn't.  AutoCAD is only three years older than Altium (counting Protel), and while I expect it has a much larger user base, it's been a mature, solid program for a LONG time.  Altium has a number of bugs that are still active in bugcrunch because they're still in the application, for over a decade.  Now I don't spend as much time in Altium as a lot of people here do, but I've put a fair number of hours into it.  I still run into situations where something doesn't work the way I expect, so then I have to play the am-I-being-dumb-or-is-this-a-bug game and most of the time it's a fucking bug.  Then I get to decide if I want to take the time to see if it's already in bugcrunch, and if not, spend more time documenting it to add it, and in the process weighing whether THIS bug is going to get enough support from other users, attract the interest of one of the devs, align with the correct phase of the moon, or whatever other factors affect the Ouija board that drives Altium's development planning.  It's a deeply frustrating and largely futile time sink, but I usually report bugs because I want Altium to be better.  I want it to keep getting better to the point where I feel like it's worth spending any money at all on the annual subscription, let along the two fucking thousand dollars they want for....what?  Interactive routing has gotten worse in recent versions! That's like THE THING that Altium is for!  How does that happen?!  Now granted, Altium is certainly has a larger and more complex data model to deal with, and that's hard, I get that.  But the sum total of my experience with it is that Altium, the company, is not really serious about making a good tool.  Dave mentioned (in this thread maybe?) that we'd "be surprised" how little effort they're putting into PCB, but at this point I see no evidence that they're putting ANY effort into it.  It's not good enough for them to be complacent, they haven't finished the job yet, and parts that they have finished, they've managed to fuck up again.

Phew, that was a long rant.  I just finished my AutoCAD project today so I can't wait to start my next project that'll be a few solid days in Altium lol.

* Not quite as absurd as in Haas CNC machines, where I have personally had the control tell me that a memory value it displays as "0.000000" (six decimal places) is not equal to "0". As if writing G-code macros didn't suck enough to begin with.

[nigelwright7557]

Given the number of free packages around now and the number of vendors too I dont think PCBCAD software is the thing to be in any more.

I used to sell it myself on ebay for about £30.
Then some one came along selling it for £4 and blew me out of the water.
Then next thing people are all using kicad so that was the end of it.

So moved on to hardware with software systems where it wont be given away.
Let the CAD vendors fight over what few customers there are left.
Best out of it.

[PlainName]

Best out of it

I'm sure more than a few here would agree with you.

[nigelwright7557]

Best out of it

I'm sure more than a few here would agree with you.

This is the reason why Microsoft systems (Windows etc) are so bloated and complicated.
Its to stop other companies competing.
Having said that sometimes a competitor pops out of the woodwork like Linux.

[tooki]

This is the reason why Microsoft systems (Windows etc) are so bloated and complicated.
Its to stop other companies competing.
Having said that sometimes a competitor pops out of the woodwork like Linux.

Oof. A lot to unpack here, since it’s all wrong.

Much of the “bloat” and “complication” are there to ensure backward compatibility with existing software and systems, something which Microsoft generally places an extremely high value on, because it’s something enterprise customers care about deeply. (Microsoft arguably takes backward compatibility to an unhealthy extreme.)

A lot of other “bloat” is to support things that are necessary. Just because you don’t need a feature doesn’t mean that having it is “bloat”. As a really simple example, text handling code, and the fonts we use, have become significantly “heavier” over the past 30 years. But it’s for a really good reason: properly supporting languages that aren’t covered in 7-bit ASCII. (The old solution of ASCII code pages sucked.) Unicode is much, much harder to support correctly, requiring a lot of code to cover how to handle it. For English, there’s little obvious value. For other languages, the benefits are immense.

Linux didn’t “pop out of the woodwork”, it was the alignment of the stars of a good open source kernel with the already mature GNU toolset. And frankly, Linux has hardly made a dent in Windows as a desktop OS, despite years of it being “ready for the desktop, for real this time!”. (The Mac has, however, made significant inroads over the past decade.) So whether it’s even truly a competitor is debatable IMHO.

Web servers were never predominantly Windows based — Apache was the leader for the longest time, now nginx is top, and Microsoft’s IIS (whose market share peaked in 2007) has dwindled to just a few percent. Linux was never really the “newcomer” in the web server market, it’s essentially always been the leader.

[T3sl4co1l]

Linux didn’t “pop out of the woodwork”, it was the alignment of the stars of a good open source kernel with the already mature GNU toolset. And frankly, Linux has hardly made a dent in Windows as a desktop OS, despite years of it being “ready for the desktop, for real this time!”. (The Mac has, however, made significant inroads over the past decade.) So whether it’s even truly a competitor is debatable IMHO.

It's not; and, as far as I can tell, never will be.  Over the weekend, I decided to dedicate a day or two puttering around with an old laptop I have that dual-boots into Debian, 9 Stretch I think.  Some casual observations:
- There's a lot of gibberish scrolling by during startup.  Am I supposed to be reading this?  I see occasional warnings, missing stuff, etc.  If there's a problem, how would I diagnose it?
- This thing hasn't booted in forever, I'm sure it's in need of updates.  Hmm, nothing seems to be going automatically though.  And I don't see a "update OS" button in the settings manager, or whatever I can tell are the nearest equivalents of that.  Matter of fact, I don't see an "update" on ANY application at all.  How, uh... just how secure is this supposed to be if there's no updates for anything, ever, anywhere..?
- I'd like to do some simple embedded dev on this thing.  Okay, so I need my toolchain.  Code::Blocks is cross platform, no problem there.  I guess it's not available via package manager, so download it.
- The files don't... "run".  They're just .deb archives.  What do?
- Okay, so almost everything pipes through apt-get, or similar package managers.  Oh also dpkg is what runs the .deb's.  Surely -- surely -- there would be a simple GUI wrapper for these, that's obviously titled, and prominently placed on the system menu?  I guess there's reason why xterm is prominently placed, but... c'mon... seriously?  Fuck is the point of booting up X if I have to memorize every magic incantation anyway?

And none of which are discoverable or obvious, so it's all searching for what to do.  Maybe there's mans on the HDD already, I don't know.  For sure, no one suggests looking at them.  Google is the holy oracle of all.  Good thing the Wifi connection "Just Works".

Discoverability is such a manifestly important goal of UX.  There isn't a single video game in the top 100 that doesn't do this.  Lead the player/user into what they need, or are most likely to need.  And automate all the piddling work for them.  (Or don't, depending on what kind of game it is; but generally, users aren't coming to an OS to play gnuClicker or whatever the equivalent would be.)

So I run some apt-get updates, upgrades, start getting some things installed that I need.  We're like 6 hours in at this point.  Having trouble locating python.  I want version 3.8 specifically, so I know it's compatible with the toolchain on my desktop.  Nope, doesn't exist, it thinks I mean some postgresql bullshit.  Nearest I can tell, the default distro does not offer specific versions.  I find it on another one (a quick search turns up dozens of sources).  I enter it by name, no good.  It does suggest adding servers to my sources.list.  Seems straightforward enough, sudo a text editor, paste it in and go.

Also, all along the way there's confusion about dependencies.  Because why would anything be easy.  Apparently you have to --fix-missing and such, and also there's a -R to install a bunch of packages in a directory in the right order.

Anyway, I finally find python and get it installing.  Text box (less).  Some deep sounding (libc6) package has a critical security update so it won't install it for me.  Or the new version is incompatible with what else I have on here.  Fine?  Q out, process finishes.  Fine, I'll just get a new version of that I guess.  Hey, why isn't sudo working anymore?  Why isn't... anything working anymore?  Reboot, maybe it's just gotten confused.  Linux is supposed to be famous for changing out whole kernel modules at run time, I don't know why this should help, but y'know, computers?  Hmm, X isn't even coming up.  It blinks a bunch then gives up and stops.  Rebooting into the "advanced" and "recovery" options doesn't make a difference.  Look up how to get a console.  Won't even login with any usernames I know.  Did... did apt-get just fucking brick the whole goddamned computer?

So I booted into Win10 the next day, and in the span of a couple hours, got most everything installed, including Code::Blocks running avr-gcc on a successful build of a recent project.  Still a slow and bothersome process, but most of it Just Works(TM).  No package hell, Installer just does the one complete configuration it's made for.  Or just dump it in some folder somewhere and run it, who cares.  Most of all, the OS doesn't let me, just, you know, fucking delete system32, as they say.


So, all this just to explain:

Even for a user who has general knowledge about computer systems,

Linux is still a piece of shit.

I'm not asking for advice; I know what I'd need to do to fix that.  Maybe it can be fixed from Windows, maybe run an installer, maybe flash a USB bootdisk.  I don't know how much that's going to wipe, if it can patch whatever fucked up in the kernel, or if I have to go through all that torture again just to get back to where I was.  I know I can look up a hundred different resources to figure that out.  Which is part of the problem by the way -- there's as many configurations as there are PCs running the stuff.  The paradox of choice.  It's a very real UX problem, this time also with very real development costs.  And with liability for the user of potentially destroying their computer.

Venting?  Sure.  Asking for sympathy?  I mean, I wouldn't mind.  Asking for solutions?  I don't care.  But that's really the root of the problem: developers write Linux for developers.  They expect all their users to know all the magic incantations, to have an internet connection always handy to search things they don't, and to just debug or recompile the kernel if anything should go wrong.  What could be easier!

It's -- it's the marketplace of ideas.  It's free software.  Free ideas.  Some people like to make a big deal about "the marketplace of ideas", but what they conveniently leave off -- or ignore with a spiritual conviction as the case may be -- ideas aren't worth anything.  In the legal sense, this is, well, patently true: you can't patent an idea.  An implementation, an invention, something physical, sure.  (Well, that's rather wishy-washy these days, between relaxed rules, less review than ever, and including more and more junk like software patents.  But, the core idea at least, of the PTO, and actual practice in recent history.)  So too it goes, free software is worth all the bits it's written on.  Free software only attains any value when it has, not just a little buy-in from users, but a truly stupendous user base -- popular projects like Firefox and Chromium are commonly touted as examples, ignoring the fact that real money supports those projects (Chromium especially, for obvious reason), and ignoring the literal millions of side projects that various users have spun off from projects big and small, as well as created themselves.  None of which you can get your hopes up about; they should be treated as only what they are: pet projects, by and for, one or a few users, with no intent of fitness-for-purpose, merchantability, support, etc.  Quite literally, you get what you paid for.

So you'll excuse me if I complain of a little cognitive dissonance, when I see optimistic claims about things people don't even realize they're extraordinarily biased towards.

Tim

[Doctorandus_P]

This Linux bashing is al lot of bollocks.

Debian is not a beginner friendly distribution. It never was, and it probably never will be.
Debian assumes you know what you're doing.
I'm not even sure if Debian "out of the box" is supposed to be used at all.
I see Debian more as a common root for more user friendly distributions.

My daily driver is Linux Mint, and it does have an update button.
Some time ago I re-started an old PC I had not used for a few years.
Just out of curiousity I tried to update with apt, and got into dependency hell.
I did not want to spend time on this, and almost wiped the SSD, then I thought of the Update Manager in Mint, and I ran it.
It solved all the dependency hell and got that old system running faultless.
It did need one or two reboots and restarts of the update manager but that's all.

About windoze...
A few years ago I was at a funeral and someone was doing a talk which got rudely interrupted because the PC which was giving a presentation on a beamer had to reboot itself because of a printer driver update.

A brother of mine had to baby sit an 6 hour CNC job because the PC wanted to reboot itself every half hour.

But in the end it's not the software which is to blame, but he people who are (ab) -using it.

And also the people who write the software in the first place.
I have not forgotten nor forgiven the FUD, smear campaigns and other anti-competitive behavior of microsoft, intel and a bunch of other "big companies"

In my end work for graduating from school I wasted several weeks because microsoft's idea of FTP was not compatible with RFC959.
It did force me to dive real deep into the FTP protocol back then (30 -odd years ago), and that was probably one of the moments that made me fall in love with Open source software.
It's not only the good they're trying to do. It's also the absence of malice which unfortunately rules in a lot of for-profit companies.

[nctnico]

This Linux bashing is al lot of bollocks.

Agreed. Every piece of software has a learning curve. Most people forget the amount of time they put into learning Windows and on top of that expect all operating systems to behave that way.

Try to switch CAD packages for example; suddenly all the 'muscle memory' is useless and the new package is total crap. Until you learn how to work with it... and only then you can have a valid opinion on which is better / worse at specific points.

[PlainName]

The thing about Windows is that there wasn't an alternative at the time. Maybe OS/2, but my experience of that was quite poor (and Microsoft deliberately broke its compatibility). Pretty much any GUI would be cool compared to a CLI. Nowadays, if you're switching away from Windows there is always the comparison with the known slick (and despite its faults, Windows is slick) example with, as you say, years or decades of muscle memory invested (although that's not translated to W10/11 so well, I think). Linux, and the Mac, have a much harder job now in converting foreign users than Windows did.

[tooki]

This Linux bashing is al lot of bollocks.

Agreed. Every piece of software has a learning curve. Most people forget the amount of time they put into learning Windows and on top of that expect all operating systems to behave that way.

Nah, the problems with Linux usability on the desktop go WAY beyond unfamiliarity. The open source model simply doesn’t lend itself towards user interface consistency, which is extremely important. This is where the Mac shines, Windows is OK, and Linux sucks.

CAD software is a genre which, regardless of platform, also generally sucks in this regard.

The thing about Windows is that there wasn't an alternative at the time.

***coughs in Macintosh***

Just what exactly do you think Windows was copying (poorly)?

Linux, and the Mac, have a much harder job now in converting foreign users than Windows did.

Having in the past sold literally hundreds of Macs to people converting from Windows (they were about 90% of our customers), it’s honestly not a huge problem for most people. The best advice I gave ”switchers” (based on what prior successful switchers had told me) was to simply “forget how you’d learned to do it on Windows, and instead just do what would make sense if you’d never used a computer”. The people who struggled were the few who expected everything to work exactly the same as Windows, which of course it doesn’t. (These were folks who had never gained any understanding of what they were doing, but rather just repeated their workflows by rote memory. I’m sure Windows 10/11 are giving them similar issues.)

Another cornerstone of a successful transition, of course, is application familiarity: Word, Excel and PowerPoint on Mac work basically the same as Word, Excel and PowerPoint on Windows. A browser is a browser. (And that already covers 90% of what most people do on computers these days…) But the same applied to users of various commercial software packages like Adobe stuff.

This doesn’t translate to Linux, because the Linux ecosystem doesn’t have that much in the way of commercial software packages, and even fewer of them are cross-platform. So users aren’t only experiencing a new OS interface, but the new-to-them open source applications, too, which may or may not even do the same things as their old one did.


People who don’t truly, thoroughly understand the ramifications of the whole user experience (the usability of the OS, usability of apps, the user interface consistency within and between apps, availability of apps, and the ecosystem more broadly) will never understand why Mac and Windows have been so successful on the desktop, and nothing else has. The typical Linux defenders show in their responses that they still don’t understand these things, which is why Linux’s desktop market share has been basically unchanged forever. (Heck, even most Windows people don’t understand them fully, which is why they never understood why Apple continued to be successful even as so-called “analysts” predicted the company’s demise.)

Did Microsoft do some unsavory things back in the day? Yes. But that really only affected Apple, and even that effect is probably not as big as people think. (I think most of Apple’s woes in the late 80s through the 90s were self-inflicted.) Nobody else was a meaningful competitor. OS/2 is the only platform I think might have stood a chance, had it not been for Microsoft’s monopolistic behaviors.


I mean, if there’s one software company that really gets me angry these days, it’s Adobe… (No, I won’t ever forgive them for buying Macromedia and then just killing off FreeHand, which was so much better than Illustrator in so many ways. But also how they used InDesign to — rightfully — lure away Quark Xpress’s entire customer base with a better product and great customer service, only to turn around once they’d gotten on top and become even more abusive than Quark had been to its customers!!)

[PlainName]

    The thing about Windows is that there wasn't an alternative at the time.

***coughs in Macintosh***

Just what exactly do you think Windows was copying (poorly)?

OK, an affordable alternative

At the time the Mac was a bit nichey. Didn't Microsoft have to bail them out at one point just so Microsoft wouldn't be a monopoly?

[tooki]

    The thing about Windows is that there wasn't an alternative at the time.

***coughs in Macintosh***

Just what exactly do you think Windows was copying (poorly)?

OK, an affordable alternative

At the time the Mac was a bit nichey.

Yes and no. Mostly yes, but at the time, they had REALLY cornered the publishing, multimedia, photography, music, and TV/film industries. Back then, you never saw a newspaper or magazine laid out on anything but Macs. Practically all professional media production was on the Mac back then. That’s really changed today, where a significant percentage of that is done on Windows.

What’s interesting to me, though, is that back then, when media production was absolutely dominated by Macs, very little of the business outside of media production was done on Macs. So you’d typically see big companies use Windows for everything, except for their art departments using Macs. Nowadays, the art departments might use Windows, but you also see lots of Macs in businesses outside of the art departments, having found their way into general office/business use a lot more. (Thanks, web/intranet apps!!)

Didn't Microsoft have to bail them out at one point just so Microsoft wouldn't be a monopoly?

Yes except actually no. Microsoft’s $150M investment in Apple was purely symbolic, to cement a big deal between the two companies: 1. a patent cross-licensing agreement (to put to bed, once and for all, the IP litigation between them), and 2. Microsoft’s promise to continue to make Mac versions of Office and Internet Explorer for 5 years. Contrary to what people often say, this investment was not needed for Apple’s survival (the company was worth billions at the time and was liquid*), and moreover, it wasn’t a cash investment, but a stock purchase. It didn’t make any tangible difference in Apple’s liquidity, but did serve to stabilize the stock price. Thus, it really cannot be called a bailout.

What’s hilarious is the way the Macworld Expo (Boston, 1997) crowd booed Bill G. when Jobs put him on the screen by satellite to announce this during the Apple keynote!

Appleinsider did a nice retrospective on it, including the video of the above: https://appleinsider.com/articles/18/08/06/august-6-1997----the-day-apple-and-microsoft-made-peace



*At the end of FY97, while the company did post a massive net loss (around $1B), the company had $4.3B in assets, including $1.2B in cash, and less than $1B of long-term debt. Heck, its market cap was just $2.3B, less than its assets! Ultimately, during the roughly 15 years (~1990-2005ish) that pundits kept predicting/announcing Apple’s imminent demise, and then once it became clear even to the dumbest of them that Apple was financially extremely stable, the few more years that they kept predicting/announcing the imminent crash of their stock price (“what goes up must come down, and it’s overpriced now!” 🤣), Apple only posted an annual net loss in two fiscal years: 1996 and 1997.

[free_electron]

one of the big problems that both linux and windows have is that they have to cater to too many hardware permutations. MacOs doesn't have that. they control the hardware and need to maintain only a limited set of drivers.

Oh , i have a 50 year old isa card for a 8 pin  dot matrix printer i want to use under windows 11. yup we have a driver for that.
oh i have a 30 year old Kodak single speed cd rom reader over a special oki plugin card .. have driver
oh, i have a 20 year old 640x480 vga webcam .. have driver

Apple goes : time to recycle your junk. here, have iForkoverhtemoney 14. its the best we made ever.

Windows 11 may solve some of it. Don't have prerequisite hardware ? sod off !
i hear you. yes you have to upgrade, yes it costs money, yes there's privacy yadda .. but if it makes the system more reliable and stable .. why are you complaining ? That was your main gripe no ? that it is unstable ?
So many bugs and security problems are caused by OLD SHIT lingering along. What was the latest one the found in linux ? a bug in a 20 year old library ? or was it 30 years old.  "oh, but you have the sourcecode" .What good is having the sourcecode if NOBODY READS IT !
they found the bug first, then traced it in the source. it's not like someone was reading the source and found it.

i have a whole library with every possible book written about electronics. i've never read any of them. i still blow up led's because i don't know what a resistor is. in fact i can't read, i just like to look at the pictures. some books i don't like cause they don't have pictures... but i still own the book so i can show off my collection to others.

THAT is the problem we are faced with. yes, apple is a walled garden but their shit works better because they control the hardware.
I'm not an apple fanboy.
- I have 3 pc's. one light for traveling , pictures etc, 2 portable workstations. all run windows 10 which can run the programs i use in life.

- i have a mac ( late 2010. faster than my latest windows laptop. try running Davinci resolve video editor. it works better on the old mac than it does on the 2 year old AMD ryzen laptop.. ). The mac is used for anything deemed safety critical : banking etc .. there is no way in hell i would access a bank website from windows.

- i have an iPhone. why ? because android is a pile of shit packaged in another pile of shit. I started with android. nothing but misery. this application can only run on that machine, this thing doesn;t scale properly, the user interface isn;t "smooth". scrolling is jerky. lollipop hasn't been ported to your hardware yet and other drivel. i had a phone ( googles red pixel first gen ) , an Asus tablet, a Coby tablet, a repackaged coby sold as honeywell for the alarm ( cause the honeywell app could only run on the modded tablet ) and a samsung tablet. nothing worked right. applications that would run on one device didn't work on another. they all ran different versions of android becasue the manufacturers couldn't update the distros in time. it all ended up in the trash can. i bought an iphone and that was it. iphone 4 was my first one. then i got an 8 , now i have an X. i am thinking about 13 but will probably wait till later this year to see what 14 brings (if it gets that 40+ megapixel camera i will bite, if not i might keep my X for another year ). i also have a first generation ipad pro.

I use windows. Why ? Because it runs the software i use and am familiar with. I Grew up with it, invested time and money learning. Not because i "like" it. An operating systems job is to let programs run. Other than that i needs to get out of my way. So far macOs and Windows cover everything i want/need. I'm done relearning everything. i want to get shit done.

My hacking days stopped with windows XP. i grew up with CP/M and dos and could write TSR's like the best. I had a licensed version of Turbo assembler. I coded an I/O driver for windows NT that ran in kernel mode so you could control home-brew hardware. It let you do peek/poke an in/out operations from userland. Written in assmbler, embedded in a turbo pascal generated DLL that could be called from visual basic , or C or whatever.

Now ? i don't care about any of that anymore. My days of making screwdrivers are over. I have a whole set. If i need a new one i will buy it.

Linux ? I Don't feel the need to learn the noble art of iron ore digging, smelting, blacksmithery to forge the steel shaft and becoming a lumberjack and woodturner to make the handle so i can lovingly create a screwdriver. ( meanwhile visiting all possible forums and having heated discussions about where to dig the best ore , what hammer to use in the forge and if oak is better than teakwood for the handle. )
I have dabbled with linux when it still came on a single 1.44Mb floppy... version 0.28 i believe. And then install 19 floppies with X-windows to drool over x-eyes . two stupid circles that followed your mouse cursor. in monochrome on 800x600 on a 386DX running at 35MHz with 2 megs of ram. It was all bedazzling as a 19 year old when windows 2.0 was top notch. Sadly i never found any software it could run. None of my dos programs could run. The same is still (partially) true in 2022. None of the windows programs run on linux (natively) or macOS. or vice versa.

Here is a question : why do the operating systems need to be so different they cant run applications ? after all it is still the same processor and the same instruction set. the CPU doesn't care.
We are spending all this effort to do the same thing , just different. how stupid is that ?

- I'll build a road here so i can drive a car to there. i built roads since the beginning of the industry. the first ones could only support pedestrians but you got there in the end. you can still walk if you want. my road is compatible with shoes.

- Some other guy goes "i don't like your road". Let me also build a road to get to the same place. it will be built differently and cars made for your road cannot run on my road natively. there is a trailer that you can put your car on that is towed by another vehicle. the trailer may break or not be able to handle the weight of your car. Drivers will constantly argue about colors , left or right hand driving , and diesel vs gasoline. Cars come with a bucket of asphalt and a manual so you can fill your own potholes. But if you alter the asphalt in any way you have to share the recipe. Others may dig up the pothole again to put their asphalt in. You can walk , or even crawl on your elbows if you wanted. Just be wary of the glass shards, splinters and ravenous bugblatter beast of traal along the way. Just close your eyes when you encounter one of those.

- A third one builds a super fancy road that can only run electric cars sold by him or a very select few. we don't want any plebeiians on our roads (and we certainly will not walk.) . have a (at extra cost of course) flower to sniff as you are being driven to your destination. Sorry, no walking here.

Meanwhile .. i want to get to that destination ! and we razed the forest to build three roads... pure idiocy

[Ed.Kloonk]

Here is a question : why do the operating systems need to be so different they cant run applications ? after all it is still the same processor and the same instruction set. the CPU doesn't care.
We are spending all this effort to do the same thing , just different. how stupid is that ?

To answer your question, you have to understand something about business law. Dos/Windows only exists because the law requires that you must make enough significant changes for software to do, yes, essentially the same thing but not so similar to be sued for copyright.

When you weigh up every good design feature and aspect about Unix and compare how it was later implemented in dos/windows you start to notice how inferior it is as a whole. And how pointless it is.

[coppice]

Here is a question : why do the operating systems need to be so different they cant run applications ? after all it is still the same processor and the same instruction set. the CPU doesn't care.
We are spending all this effort to do the same thing , just different. how stupid is that ?

The instruction set doesn't care, but everything else does. A few user->system APIs are clearly dumb, but most differences are a design choice that its advocates can make a good argument for, and detractors can make a good argument against. With "the right way" of doing things not being clear, its not appropriate to force every system to work the same way and be compatible. Fudge layers, like WINE allowing a Linux machine to run a range of Windows apps, exist, but are always a compromise. Various cross platform toolkits allow an application developer to pay very little attention to the different systems, except for the areas where the special qualities of each system make a real difference to the particular application. For most simple applications, which do nothing exotic with the system they run on, people overplay just how different the various modern systems are, and how much extra work it takes to be cross-platform. The problems usually arise from people buying into development tools that are tied to one particular system, which are then a huge PITA to port to other systems.

[nctnico]

Meanwhile .. i want to get to that destination ! and we razed the forest to build three roads... pure idiocy

No, it isn't. It is more like bycicle roads, highways and train tracks. Each type of transport needs their own type of infrastructure. Linux, OSx and Windows each cater to a different kind of use / user. They aren't interchangeable.

[Gribo]

Good luck using that 40 years old ISA card without specialized drivers & VM. Windows stopped supporting direct IO for ages. But, on the other hand, that HPIB instrument? Yep, plug it into that USB GPIB adapter, and start working.
Windows does have binary compatibility with WinAPI applications - you can run a windows 3.0 app, as long as it was properly written (no direct HW access, no DOS calls).

[PlainName]

yes you have to upgrade, yes it costs money, yes there's privacy yadda .. but if it makes the system more reliable and stable .. why are you complaining ?

Well, that's the problem. It's less reliable and stable, plus it's a worse user interface. Why on Earth would anyone willingly move to it?

[Karel]

Linux is designed by engineers for engineers.
Windows is designed for gamers and housewives.
 

[free_electron]

Linux is designed by engineers for engineers.
Windows is designed for gamers and housewives.
 

engineers. lol. so they can drive a locomotive.

[free_electron]

Here is a question : why do the operating systems need to be so different they cant run applications ? after all it is still the same processor and the same instruction set. the CPU doesn't care.
We are spending all this effort to do the same thing , just different. how stupid is that ?

The instruction set doesn't care, but everything else does. A few user->system APIs are clearly dumb, but most differences are a design choice that its advocates can make a good argument for, and detractors can make a good argument against. With "the right way" of doing things not being clear, its not appropriate to force every system to work the same way and be compatible.

that's what the BIOS used to be fore. that was your lowest level API.  Read sector, write sector , read byte, write byte.
It is essentially what Apple did with the lisa and early mac's. The hardcoded roms provided a base layer of operations .
if we had a common base layer that could do things like ' open file , read file , write file close file. give me a chunk of memory. give me a window. then the OS would just be "paint". The cores are the same.

Essentially a virtual machine that exposes a bare minimum. you can get a chunk of memory and you can get a canvas to draw on ( screen, printer , doesn't matter ) . "Drivers" would be elements that work at the VM level and translate the "canvas" to hardware ( video card ) or output ( PDF , postscript , pcl . whatnot )

You don't need a bazillion functions.
file access (open close read write create delete file/directory , select source (local/remote doesn't matter. its unified . everything accessed through GUID)
Give me window to draw on. you cannot step outside this window. base functions to draw lines and whatnot.
Programs at startup get a chunk of memory and can request/release more if needed. They cannot step outside their chunk.

Programs are installed in a single container. They don't vomit their stuff all over the place. no shared libraries. disk space is cheap. if application x needs version y of a library it is in its container. if application z needs another version : it can be found in its container.
i hear you buzzing : but but security updates and patches in libraries : up to the application manufacturer to release a new package. This approach will force them to keep things up to date.

so many times programs have quirks because some shared dll got updated and there is now some weird incompatibility.

There are operating systems that can do this. iRMX can do it. ok, now it is no longer compatible with new stuff cause it isn't being maintained but i have seen machines that ran iRMX with dos in one container, windows 2.0 in another , wind 3.11 in another and win95 in yet another. and native iRMX applications in their own container. In the waferfab we had expensive ion implantation machines that ran iRMX. When you needed to open a valve there was a button drawn on the touchscreen. that button was a small irmx application that lived in its own space and could only control that valve. that application could not step outside of its container.

If you think about it this is nothing but system virtualisation.

If you look at linux : there are what 300 different flavors of the darn thing. And what makes them different ? eye-candy ! the core functions are the same. but the eye candy difference are large enough that applications are not always runnable.
Its a headache both for users and coders as they need to maintain compatibility with all these little things.

The PC was a clever idea. here is a hardware box that has a bios. you interface with this thing to operate the machine. int21 got you almost anything you wanted.
why can't they create such a thing at a higher level. then it wouldn't matter. applications would work no matter what as they would contain everything needed.

Anyway, i'm at the point in my life where i don't care anymore. Like i said : if it gets the job done its fine with me. The less effort needed on my side the better. i don't want to constantly learn new things, debug someone's source code. if that means i need to pay Microsoft or apple 100$ for a license. so be it. Linux isn't free either. it takes time. The effort in time needed on my part is valued much more than 100$. We live a finite amount of time. i'm past my halfway point. There's places i want to see and stuff i wan to do. learning yet another os is not on my list.

[free_electron]

Meanwhile .. i want to get to that destination ! and we razed the forest to build three roads... pure idiocy

No, it isn't. It is more like bycicle roads, highways and train tracks. Each type of transport needs their own type of infrastructure. Linux, OSx and Windows each cater to a different kind of use / user. They aren't interchangeable.

you are comparing apples with pears.

i'm talking desktop and desktop only. windows, macos, linux, solaris , BeOs , whatever in all their non-server versions.
They all do similar things : they expose a filing system and abstract a bunch of hardware and provide an API to applications. And they expose an interface to the user in the form of a graphics system.

The user runs applications.
I don't give a rats ass how the "close" button is handled internally or what shape it is. as long as every applications uses the same close button. not one program that has a button top left, another that has a button bottom right. uniformity. file open menu is the same for every applications. Learn it once.

I want to do work. work for which i need applications. applications that run on hardware . that "operating system" is only an abstraction layer between hardware and applications. It should be invisible to the user.
Now there are different abstraction layers that are incompatible with each other. the whole point of an abstraction layer is to make things uniform.

I find this weird. just me being me i guess.

[nctnico]

Meanwhile .. i want to get to that destination ! and we razed the forest to build three roads... pure idiocy

No, it isn't. It is more like bycicle roads, highways and train tracks. Each type of transport needs their own type of infrastructure. Linux, OSx and Windows each cater to a different kind of use / user. They aren't interchangeable.

you are comparing apples with pears.

No, not at all. Forget about desktop versus server. I spend quite some time developing embedded software which I really wouldn't want to do on Windows. First of all many of the targets I work with run Linux and secondly running a compiler on Linux compiles software a whole lot faster than Windows. A decent sized embedded microcontroller project takes minutes to compile on Windows versus tens of seconds on Linux (using the same PC, same compiler and same number of parallel threads). And there are many other computing tasks where Linux works much better for me compared to using Windows.  Windows simply isn't the right tool for the job I do because it gets in the way of productivity.

[free_electron]

Meanwhile .. i want to get to that destination ! and we razed the forest to build three roads... pure idiocy

No, it isn't. It is more like bycicle roads, highways and train tracks. Each type of transport needs their own type of infrastructure. Linux, OSx and Windows each cater to a different kind of use / user. They aren't interchangeable.

you are comparing apples with pears.

No, not at all. Forget about desktop versus server. I spend quite some time developing embedded software which I really wouldn't want to do on Windows. First of all many of the targets I work with run Linux and secondly running a compiler on Linux compiles software a whole lot faster than Windows. A decent sized embedded microcontroller project takes minutes to compile on Windows versus tens of seconds on Linux (using the same PC, same compiler and same number of parallel threads). And there are many other computing tasks where Linux works much better for me compared to using Windows.  Windows simply isn't the right tool for the job I do because it gets in the way of productivity.

so linux plays well when linux is the target. got it.
Code writing is only one thing.

Let me give you an example of what i mean with "stupidity".

I need to get a document form my mortgage company to give to my tax preparer.
fire up a web browser , accept cookies ... download the document . it save this in the downloads folder as q573612_44837.pdf
double click the document to open in acrobat. acrobat wants to update. check document is filled out properly , save document as mortgage_statement_2021.pdf , throw it in "tax documents 2021" folder.
go back to browser. access tax preparer website. find upload button , go find the damn file ( by default it opens on to desktop , which is not where i stored the file) wait fo r it to upload.

why can i not just speak into the microphone like in start trek ? computer, get the mortgage tax document from the bank and send it to luke at the tax preparation office. save a copy in my tax 2012 folder.
Many programs these days use different file selection interfaces. We have a whole range of web browser out there that all use webkit as a core. why ?
can we standardize this stuff so we don;t have to rel-learn it every time.

ApplE fucked up big time in their last update. the search bar in the browser on an iphone was always at the top. now it is at the bottom. this irritates me to no end. i can't even be bothered to find out how to move it. i prefer being irritated over its position than being irritated on how to figure out to move it.

user interfaces these days SUCK! i had a rental car two weeks ago. a KIA niro. nice car. but the user interface ? my god what a button fest ... the steering wheel had like 30 buttons, 4 scroll / click / sideclick wheel. The two control stalks had so many functions they ran out of space to put texts on them to tell you what they did. The touchschreen has soft buttons to select radio , camera, gps AND hard buttons underneath that do EXACTLY THE SAME FUNCTION as the touch zones on the display ! what drunken idiot designed that ?
And then the GPS insisted i enter a destination by city first, then street, then number. i don't know any of that. i needed the Hilton hotel. Just let me scroll the map from where i am. I know its the closest one to the airport here. why can't i just type in  "Hilton" ?

now, back to your compiler. I really doubt this is an operating system issue. this is a compiler issue. You say it the same compiler runs on windows and linux. that in itself is impossible. linux cannot run windows binaries and vice versa. It may be the same "product" from the same "company" , but it is NOT the same binary. if it runs that much slower on one OS versus another there is a problem in the application. It's more optimized for one platform than another.

what does that compiler do ? take a chunk of text and translate it into opcodes. This is all done in memory with some disk access for files. You can't tell me that linux is that much faster for memory allocation/deallocation and disk i/o than windows. Differences , yes. a few percent here and there. Same for threading. That is not possible. but minutes to tens of seconds ? that is not the OS fault. that is an application issue. it is just not written right for that OS.  Or they use an emulation layer. This thing is developed natively for linux API. it runs on windows using a Linux to windows translation api. it is not native to windows. This leads me back to my earlier statement : why don't we have a common API that every OS uses. Actually we would not need an 'operating system' as such.

After all .. what is an operating system ? a core that does file and memory i/o and graphics. anything else is an application.
file browser ? application.
web browser ? application .
printer control panel ? application

The core should handle connectivity and device access in a standardized way. applications don't get to do that. There would be one core and one core only. I like the clipboard program from suse, the printer finder from microsoft , the file browser from apple. They are all compiled for a PC so they can run.

There would be no duplication of effort and we could select the bits and pieces we liked. Competition would be purely on the abilities of the applications (for software) and the speed of the hardware (for hardware vendors.). Now we're stuck in a situation where i need a windows machine for this, a linux machine for that and an apple box for something else.
Applications should run in single containers. Each app has its own world.. if i buy a new computer : move the application container to the new hardware. no more need for installers. an application should be a single file with an embedded file system. myprogram.app -> an encapsulated file system that contains everything needed to run this thing. user settings are saved to this container as well. or you could do myprogram.settings.<user>

Applications do not get to paint their own buttons and user interface elements. those are handled by the UI. the application requests a viewport ,in this vieport there is a button x long by y high with the text "Ok" in it. the UI application decides how it is actually painted, font and whatnot. Enforced uniformity.

if i look at my pc: I use a certain look and feel, but many programs bring their own look and button style . Some even have different looking close buttons. this is madness. the computer is set up with a specific color scheme that suits me (i'm colorblind so i use my own color scheme) and font size and whatnot. Half of the programs bring their own look and feel that is so far off and not adjustable to this UI style. It is irritating ! One program uses this font as default in that size, another uses another font in a different size. one is dark mode, the other is light mode . These things need to be global settings with no overriding ability from inside an application.

Now every time i buy a new computer i spend hours loading it with everything i want. in this case : just copy the .app files over. everything travels with it. settings and all.

i'm amazed that in 2022 we still have to put in so much effort for things that can be solved in a simple way. In fact, these things are not new. it used to be like that under dos and cp/m and many older 'systems'. zip the folder , copy unzip. done.

[tooki]

You don't need a bazillion functions.
file access (open close read write create delete file/directory , select source (local/remote doesn't matter. its unified . everything accessed through GUID)
Give me window to draw on. you cannot step outside this window. base functions to draw lines and whatnot.
Programs at startup get a chunk of memory and can request/release more if needed. They cannot step outside their chunk.

You don’t have the foggiest notion what operating systems actually do, then. If our operating systems did only what you propose (ignoring the fact that there are many, many ways to do those same things), we’d be going back to the operating systems of the 1960s-70s.

Networking as we know it? Gone. Internet? Gone. Drag and drop? Gone. A desktop? Gone. Standardized UI elements? Gone. Printing anything but monospace text? Gone. Sound? Gone. Inexpensive software? Gone. Peripheral support? Gone. Those things all work BECAUSE we have these “fat” operating systems running on top of the kernel, which is basically all you’ve proposed. Those fat layers make it possible to develop software quickly (and thus cheaply), have compatibility between programs, and have greater software stability (and usability) because developers aren’t wasting their time reinventing the wheel over and over again.

So yes, our OSes have a “bazillion” functions and that’s a decidedly good thing!!!!

[T3sl4co1l]

Talk about missing the point...

I got the impression from that comment that, basically we could return to the days of the "dumb terminal", but a bit smarter.  I mean, there's still a few smart terminals these days, netbooks, etc.  But from a somewhat different tack: there's quite some justification for doing it this way:
- What with how incredibly powerful browsers are these days;
- And consider the walled gardens (albeit they're low walls, depending) of say mobile phone ecosystems, with an app store to select software, which doesn't even run natively on the machine anyway;
- We could just make the browser the OS instead.

It's like... shades of MS integrating IE into Win9x, but going entirely the other way around, integrating an OS into the browser.  Right?

You've got free directory services.  Just view file:/// as FTP directories from browser-internal templates.  Add some POST functionality, drag-and-drop actions, etc. and there you go, a fully functional file manager.  Make local pages a bit richer in permissions, and you've got file system protections, XSS protection, etc.

Layout of apps is relatively trivial; no need for layout managers, specify everything in CSS/HTML/JS.  Maybe you'd want something a bit tighter than JS, fine, but also JS is the de facto web standard so you're going to have a damn hard time removing it.  Or like how e.g. Android stuff is built on Java (as I understand it??), so it's not even native, it's JIT or interpreted (as is JS).

No shortage of languages and development possible.  If we include WebAssembly, it'll handle basically any binary compatible with that interface; lots of things can be transpiled to JS, Java or WA; and in the worst case, things can be compiled to ARM/x86 and virtualized/emulated as needed.

And so you can very well have something like ye olde BASIC ROM, but instead of LOAD "*",8,1 it's a proper graphical interface with both a local file system and internet connectivity; and instead of the lingua franca being BASIC, it's the richness that we expect and demand in modern times: a fully featured JS, or Java, or Python, or whatever console/environment you like.

Which, to bring it back a bit -- the mention of having a canvas to draw on and do whatever with, is basically this (plus reading into it a bit, in terms of supporting features).  Modern OSs, and the web services they run in turn, are modeled on fairly standard visual objects, -- object orientation at that -- and built-in compositing.  You don't need to draw a fuckin' combo box, you can instantiate one and let the OS handle it.

So I find it very believable that we could have an OS, which does basically the things we expect of it, and that is divorced from much of the historical baggage, the questionable backwards compatibilities that all of them suffer (Win/DOS and *nix); and that supports, at a fairly basic level, all the rich features we expect -- and no more, and no less, anything extra being defined by apps.

And that's a pipe dream of course; such a thing will not gain widespread adoption, not very quickly at least.  But as dreaming of pipes go -- it's one of the more believable ones, and I can see how it might go.

I'm not too familiar with Android, or iOS for that matter, how they're structured (aside from a linux-y core?) and what they can do; it seems to me, there's about four major things missing here:

Input - KB&M support, and just generally all the things we expect to be able to use.  Game controllers, but also more special-purpose devices.  Mobile devices are optimized for touch, which is neither very rich nor precise.  Some phones do let you use KB/M via USB/BT, but app support probably isn't great..?  Conversely, PC apps are rarely(?) made for touch, so, point in case, there's a big divide between these extremes.

Output - big monitors, multiple monitors, more powerful GPUs (granted, they're startlingly powerful already these days), just, more IO in general -- just, bulkier things, expansion slots over USB ports, say.  It's hard to imagine a cell phone plugged into a 2k+ display, let alone typical apps being any kind of useful at such scale and resolution.  And, like -- networking, hard drives, etc. can be connected by radio, sure, even at quite competitive bandwidths these days, but there's something to be said for wired connections -- no need to pollute the airwaves, or risk potential snooping of your data in transit.

Purpose - mobile devices are hand-held, you can't exactly do dev work on them, or much of anything professional aside from showing off or taking pix/vids, and making phone calls.  Arguably driven by the above point, but arguably the converse is true, that interfaces are driven by purpose instead, too!

Security - mobile devices are relatively locked down, for several good (and not so good) reasons.  The imagined OS should be more flexible to support a more desktop-like model, while still observing important isolation and security features.  Which might be a good justification to not actually offer truly universal drag-and-drop (as has been the dream since the 60s, alas).

Or about the example of narrating one's intent -- even given a ~100% accurate speech interpreter, the computer needs to know that you're even authorized to do those things, let alone to know how to do all of them when none of those documents (randomly named tax doc scan, tax doc submission field, etc.) are tagged as such.  (Such schemas might exist -- although the closest I see on schema.org offhand is taxID, hardly descriptive enough here.  Probably one would have to consult a much richer knowledge system to have enough information types to be able to handle these sorts of things -- and in turn, the ability to parse and recognize a document as being of such-and-such a type, and the webpage offering/being capable of receiving such data as well.  Which, I mean, it's a big hurdle, but it's also one that's fairly well explored these days, and could be harnessed to facilitate applications like this.)

Tim

[free_electron]

You don't need a bazillion functions.
file access (open close read write create delete file/directory , select source (local/remote doesn't matter. its unified . everything accessed through GUID)
Give me window to draw on. you cannot step outside this window. base functions to draw lines and whatnot.
Programs at startup get a chunk of memory and can request/release more if needed. They cannot step outside their chunk.

You don’t have the foggiest notion what operating systems actually do, then. If our operating systems did only what you propose (ignoring the fact that there are many, many ways to do those same things), we’d be going back to the operating systems of the 1960s-70s.

Networking as we know it? Gone. Internet? Gone. Drag and drop? Gone. A desktop? Gone. Standardized UI elements? Gone. Printing anything but monospace text? Gone. Sound? Gone. Inexpensive software? Gone. Peripheral support? Gone. Those things all work BECAUSE we have these “fat” operating systems running on top of the kernel, which is basically all you’ve proposed. Those fat layers make it possible to develop software quickly (and thus cheaply), have compatibility between programs, and have greater software stability (and usability) because developers aren’t wasting their time reinventing the wheel over and over again.

So yes, our OSes have a “bazillion” functions and that’s a decidedly good thing!!!!

not true. anything you have described are applications.
- networking ? nothing but remote data access. data that resides in remote storage, is read , parsed and visualised or stored on local storage. Hey operating system , fetch me index.html at www.google.com
- internet ? see above. nothing but remote file reading
- drag and drop ? nothing but an application that tells the os to grab file x from location y and move it to location z. drag and drop for other objects ? window manager telling application object x has moved from location y to location z. do something.
- desktop ? no difference.
- standardized ui elements : part of the os window manager.

you are not sacrificing anything.
the problem is that today we have multiple of these 'standard' systems and they are all incompatible with each other !
it would be great if we could make applications that could completely run inside a browser. then the browser becomes the OS layer. at least that world is (sort of) standardised. doesn't matter if you us chrome , safari , or edge. the site works.

That is what i am talking about. i can pick the os with the look and feel i want and can run any application. applications are not tied to one OS.

[free_electron]

- We could just make the browser the OS instead.

exactly !. there are whole complex cad systems that run in-browser. doesn't matter if you are on mac, windows or linux or even solaris. have a modern browser ? it'll work. the hardware becomes abstract. the OS disappears.  a computer can run any application, irrespective of who made the hardware and the controlling ... eh should we call it firmware ? So now the whole ecosystem changes. The endless x better than y is purely based on speed of the box and the built in peripherals it brings.  the programs will run. the "computer" is now defined purely by the cpu speed, storage size , screen size etc.

[nctnico]

Talk about missing the point...

I got the impression from that comment that, basically we could return to the days of the "dumb terminal", but a bit smarter.  I mean, there's still a few smart terminals these days, netbooks, etc.  But from a somewhat different tack: there's quite some justification for doing it this way:
- What with how incredibly powerful browsers are these days;
- And consider the walled gardens (albeit they're low walls, depending) of say mobile phone ecosystems, with an app store to select software, which doesn't even run natively on the machine anyway;
- We could just make the browser the OS instead.

Something along that idea exists for a long time already (decades!). It is not called a browser but a graphical terminal. And it is used on a large scale even today although the name terminal has been replaced by 'thin client'. These are typically used where people need some custom application for data entry / reading and a glorified typewriter.

Additionally Google and Microsoft (AFAIK) also provide webbrowser-based document editing & sharing.

[tooki]

You don't need a bazillion functions.
file access (open close read write create delete file/directory , select source (local/remote doesn't matter. its unified . everything accessed through GUID)
Give me window to draw on. you cannot step outside this window. base functions to draw lines and whatnot.
Programs at startup get a chunk of memory and can request/release more if needed. They cannot step outside their chunk.

You don’t have the foggiest notion what operating systems actually do, then. If our operating systems did only what you propose (ignoring the fact that there are many, many ways to do those same things), we’d be going back to the operating systems of the 1960s-70s.

Networking as we know it? Gone. Internet? Gone. Drag and drop? Gone. A desktop? Gone. Standardized UI elements? Gone. Printing anything but monospace text? Gone. Sound? Gone. Inexpensive software? Gone. Peripheral support? Gone. Those things all work BECAUSE we have these “fat” operating systems running on top of the kernel, which is basically all you’ve proposed. Those fat layers make it possible to develop software quickly (and thus cheaply), have compatibility between programs, and have greater software stability (and usability) because developers aren’t wasting their time reinventing the wheel over and over again.

So yes, our OSes have a “bazillion” functions and that’s a decidedly good thing!!!!

not true. anything you have described are applications.
- networking ? nothing but remote data access. data that resides in remote storage, is read , parsed and visualised or stored on local storage. Hey operating system , fetch me index.html at www.google.com
- internet ? see above. nothing but remote file reading
- drag and drop ? nothing but an application that tells the os to grab file x from location y and move it to location z. drag and drop for other objects ? window manager telling application object x has moved from location y to location z. do something.
- desktop ? no difference.
- standardized ui elements : part of the os window manager.

you are not sacrificing anything.
the problem is that today we have multiple of these 'standard' systems and they are all incompatible with each other !
it would be great if we could make applications that could completely run inside a browser. then the browser becomes the OS layer. at least that world is (sort of) standardised. doesn't matter if you us chrome , safari , or edge. the site works.

That is what i am talking about. i can pick the os with the look and feel i want and can run any application. applications are not tied to one OS.

Just more confirmation that you do not know what OSes do and how they work.

No, networking is not just remote file fetch.

No, that’s not all drag and drop does.

You didn’t mention the window manager having a UI widget toolbox. You just said a blank window to draw lines in.

We already can run apps completely within a browser.

And finally, a common baseline means making HUGE sacrifices. It means devs are limited to the specific API calls common to all OSes. This is precisely why present-day cross-platform frameworks suck: you can’t use anything that isn’t present on ALL supported OSes. The Mac, for example, has a rich set of well-designed UI elements of which numerous are not available on other OSes.

So sorry, you don’t know what you’re talking about.

[tooki]

Something along that idea exists for a long time already (decades!). It is not called a browser but a graphical terminal. And it is used on a large scale even today although the name terminal has been replaced by 'thin client'. These are typically used where people need some custom application for data entry / reading and a glorified typewriter.

Additionally Google and Microsoft (AFAIK) also provide webbrowser-based document editing & sharing.

Google, Microsoft, and Apple (and numerous others) all have web based office software suites.

The problem is that they don’t perform quite as well as desktop apps, in particular with proper OS integration (like file association).

[MadScientist]

You don't need a bazillion functions.
file access (open close read write create delete file/directory , select source (local/remote doesn't matter. its unified . everything accessed through GUID)
Give me window to draw on. you cannot step outside this window. base functions to draw lines and whatnot.
Programs at startup get a chunk of memory and can request/release more if needed. They cannot step outside their chunk.

You don’t have the foggiest notion what operating systems actually do, then. If our operating systems did only what you propose (ignoring the fact that there are many, many ways to do those same things), we’d be going back to the operating systems of the 1960s-70s.

Networking as we know it? Gone. Internet? Gone. Drag and drop? Gone. A desktop? Gone. Standardized UI elements? Gone. Printing anything but monospace text? Gone. Sound? Gone. Inexpensive software? Gone. Peripheral support? Gone. Those things all work BECAUSE we have these “fat” operating systems running on top of the kernel, which is basically all you’ve proposed. Those fat layers make it possible to develop software quickly (and thus cheaply), have compatibility between programs, and have greater software stability (and usability) because developers aren’t wasting their time reinventing the wheel over and over again.

So yes, our OSes have a “bazillion” functions and that’s a decidedly good thing!!!!

not true. anything you have described are applications.
- networking ? nothing but remote data access. data that resides in remote storage, is read , parsed and visualised or stored on local storage. Hey operating system , fetch me index.html at www.google.com
- internet ? see above. nothing but remote file reading
- drag and drop ? nothing but an application that tells the os to grab file x from location y and move it to location z. drag and drop for other objects ? window manager telling application object x has moved from location y to location z. do something.
- desktop ? no difference.
- standardized ui elements : part of the os window manager.

you are not sacrificing anything.
the problem is that today we have multiple of these 'standard' systems and they are all incompatible with each other !
it would be great if we could make applications that could completely run inside a browser. then the browser becomes the OS layer. at least that world is (sort of) standardised. doesn't matter if you us chrome , safari , or edge. the site works.

That is what i am talking about. i can pick the os with the look and feel i want and can run any application. applications are not tied to one OS.

You are either deliberately mis representing things or simply actually haven’t a clue how modern OS and GUIs work. Drag and drop is not about files that’s merely one aspect of it. Internet access is not about just about files

You are presenting a 1980s argument in 2022 , OS functionality is there to present abstractions for data handling , interprocess/inter task management and communications , data object interchange , multi media interchange etc etc etc. The alternative is each application has to build all this stuff itself. ( which was happened in the 80s) and inevitability there is then no ability to share anything

However an modern OS designer has to pick a particular method of implementing all this functionality and as a result we have different ways of achieving that end.

Whinging about “ why can’t I do this “ is just nonsense. We are where we are because the PC market has competition ( however flawed ) and in a competitive environment you will always have someone saying “ look here I have a better way to do it “

[free_electron]

You are either deliberately mis representing things or simply actually haven’t a clue how modern OS and GUIs work. Drag and drop is not about files that’s merely one aspect of it. Internet access is not about just about files

so then tell me : what is missing ?
You keep on sticking to existing methods and stuff. raze it !.
Make everything a storage medium that stores files. Every file has privileges. Every file can be accessed using a global indexing system.
no more need for complex systems. it's all transported using one method.

web server ? don't need them. simply access index.html on volume www.google.com .  A web server becomes nothing more than a remote file store. the URL type tells you what it is.
your own local harddisk is drive_c@.mycomputer or drive_d@mycomputer ( or something along those lines. ). A computer that is network connected gets a sort of DNS entry and can be connected to. ( there would be local hostnames too of course similar to 127.0.0.1 . essentially .mycomputer is "home")

www.google.com -> remote web server. a web browser will read index.html , parse and do the rest.
<anything>.mycomputer -> local stuff . drive_c.mycomputer , epson7100.mycomputer . this is already what unix systems do for devices.
<somebody>@mailservice.com -> filestore for email purposes.
remote applications run as a service that transport packets of data ( a packet is nothing but a small file if you think about it) , parse process and respond with other files.


The network stack could be much simpler now. no more need for all those different protocols. it's a simple file transport . email ? same thing. an email is nothing but a small file that gets sent and received to a target. a storage bin somewhere with a public alias. me@mailbox.com is nothing but a remote drive that holds the files until you pick em up. if i send an email to me@mailbox.com all i do is write a new "file" to that storage bin. that file can have a name like <sender>_<timestamp>_guid.mail.

everything is a simple file and file read / write operation at "global level".
On your hardware there is a driver that lets the operating system read/write files over network and store them on local volumes.
you would need minimal firmware to do this. you don't need random file access : bulk transport. 99% of cases covered. remote databases would use a soap like mechanism to access. commands and control files and data bundel files coming back. it doesn't stop anything.
you'd have a lightweight networking and storage kernel.

interprocess communication ? concept of a file in ram.
the actual operating memory would also be like a file. when a program is loaded from harddisk a single file ( the container for the program ) is read into a memory file. think of the computer memory as a huge ramdrive. programs that are allocating memory are doing nothing but creating a larger file. runtime memory is a ramdrive. run out of ram : swap it to disk. one program has one runtime file. can't step out of boundaries. the operating system task is to allocate , manage and abstract the physical storage into a single file.

everyting else are applications that do stuff with the contents of the files.

The windowing manager provides a common set of UI elements and a drawing canvas. need anything not in the default ui element stack ? you can make your own. there are drawing commands and event messages.

[cuote] in a competitive environment you will always have someone saying “ look here I have a better way to do it “
[/quote]
absolutely ! i wan the best of the best. but now i need three or more operating systems to satisfy that !. x is the best in class but windows only , y is best in class but mac only. z is best in class but linux only ... with such a system it would not matter anymore. i can run anything.
i simply buy my "base system" (hardware + the base kernel) purely on the speed/memory and peripherals i need. if apple makes a faster box than dell  or lenovo has the peripherals i want i buy that box. the applications will work.
operating systems would purely compete in implementation. the best operating system is the one that can get the work done the fastest with the least amount of resources. The applications would run no matter who built the OS.
You could extend this to hardware to a point. optimize the motherboard and base pack. we use standardised i/o : ethernet and usb. actually, ditch ethernet. do everything over usb. wifi ? usb based adapter. classic ethernet ? usb based adapter. usb again is simplified to do "file" transport. usb is already point to point.

when i mean "file" i mean a packet of data from beginning to end.
there would be one other interface for keyboard/mouse and other "slow user input". that works with very small , rapid burst packets. ( single keystroke, mouse move, gamecontroller move ). something like a packet containing 8 bytes or 10 bytes. always fixed layout.
all the rest has intelligence in the adapter.

fun to think about. There have been systems that worked like that...

how do web applications work ? its all html transport.

[nctnico]

The network stack could be much simpler now. no more need for all those different protocols. it's a simple file transport . email ? same thing. an email is nothing but a small file that gets sent and received to a target. a storage bin somewhere with a public alias. me@mailbox.com is nothing but a remote drive that holds the files until you pick em up. if i send an email to me@mailbox.com all i do is write a new "file" to that storage bin. that file can have a name like <sender>_<timestamp>_guid.mail.

No, like others already wrote: your view is way too simplistic. I just picked this snippet as an example to show that your suggestion is not how it is done. Email is typically (*) stored in a database so it can be indexed (and easely transported from one computer to the other when replacing a computer). Storing every single email in a single file is prone to trouble and cumbersome to index. Back in 1995 I used Eudora and that already used a database-ish system with indexes to store email.

* Some email programs do use files but that dates back a few decades when email clients like Pine and Elm where used on Unix systesm. That is when internet was in its infancy and webbrowsers had not been invented yet.

how do web applications work ? its all html transport.

Not quite. It is http transport but the contents can be anything. Like JSON or XML formatted database records to feed data into an HTML5 or Javascript application running in the client's side browser.

[MadScientist]

Sophisticated web applications today are combinations of back end server processing and front end client processing , it’s much more then passing index.html around 

Interprocess communications needs queues , locks, semaphores, mutexs, all supplied by the OS.

The whole thing is simply way more complex then you present it , your perspective is DOS from the 70s , where the OS provides extremely primitive support and the application must do everything else.

This leads to bloated apps , vast differences in user interfaces , dont you remember the pre gui applications , everything was different , nothing was interoperable , it’s was a nightmare

[tooki]

You are either deliberately mis representing things or simply actually haven’t a clue how modern OS and GUIs work. Drag and drop is not about files that’s merely one aspect of it. Internet access is not about just about files

Please learn to quote correctly. You’re attributing to me something said by MadScientist.

so then tell  me : what is missing ?

Where to begin…

No, seriously, you have NO clue how software works, and why your proposed system can’t work. There’s a good reason we abandoned that approach long ago: it’s inefficient, error prone, and simply not up to the task of the things we use computers for these days.

You keep on sticking to existing methods and stuff. raze it !.
Make everything a storage medium that stores files. Every file has privileges. Every file can be accessed using a global indexing system.

That is how some aspects of modern OSes already work and have for decades.

But you clearly haven’t the foggiest notion why a simple “everything is storage” model is not a viable approach. But suffice it to say that it’s not, which is why we use different mechanisms for different things.

Not that I think you understand file handling anyway.

[mountains of drivel]
how do web applications work ? its all html transport.

Oh sweet summer child… 1. It’s not all HTML. 2. It’s not all HTTP transport. (In fact nowadays most of it isn’t HTTP, it’s HTTPS at minimum, but that’s far from the only protocol.)


The complexity in modern OSes is there for a reason: it takes care of the complexity ONCE so that application developers don’t have to reinvent the wheel over and over again. That’s how things used to work, and it sucked.