KiCad 6 is coming! (has arrived!!)

[Cerebus]

Why, if one was a Linux user, would one choose a proprietary hypervisor when there are perfectly good free ones? The only reason to be running VMware workstation under Linux that I can see is if one wanted to maintain integration/compatibility with a VMware ESXi estate, or if VMware is all that one understood how to 'drive'. (I've nothing against VMware itself, I use VMware fusion myself on macOS, but I do have ESXi machines to integrate with.)

This is a bit off-topic but what VMware can and Virtualbox does not is being able to run 3D accelleration in a virtual machine. Or better put: all my attempts to use 3D (like OpenGl) in Virtualbox has failed where with VMware it worked right out of the box.

I've had Altium working "well enough" in a VirtualBox VM on Linux and macOS, but I'd agree one's chances of first time success are higher with VMWare.

[chickenHeadKnob]

Why, if one was a Linux user, would one choose a proprietary hypervisor when there are perfectly good free ones? The only reason to be running VMware workstation under Linux that I can see is if one wanted to maintain integration/compatibility with a VMware ESXi estate, or if VMware is all that one understood how to 'drive'. (I've nothing against VMware itself, I use VMware fusion myself on macOS, but I do have ESXi machines to integrate with.)

This is a bit off-topic but what VMware can and Virtualbox does not is being able to run 3D accelleration in a virtual machine. Or better put: all my attempts to use 3D (like OpenGl) in Virtualbox has failed where with VMware it worked right out of the box.

I've had Altium working "well enough" in a VirtualBox VM on Linux and macOS, but I'd agree one's chances of first time success are higher with VMWare.

A few pages back I complained that I couldn't get the 6.0 nightlies running in virtualbox mint linux 20.2. What was happening (I think) was that eeschema,PCBnew ect were seg-faulting on graphics. The only reason I was trying this on VirtualBox was that I was loath to upgrade my base system (Mint 18.x) as I had invested a lot of time tweaking everything. Instead of diagnosing further I bit the bullet and shit-canned the whole thing and did a clean install of Mint 20.2 and natively run 6.0 official which works fine so far. The whole upgrade of my system was a big dick-around I wanted to avoid but that has nothing to do with Kicad.

[holgerinput]

Python 3.9, which itself doesn't support Windows 7 anymore.

I haven't looked into that yet, but my StreamDeck doesn't support Windows 7 in that it won't install, and if you extract the files it won't run because some DLLs are missing. The trick there is to shove fake DLLs in the install directory, and it runs. The only purpose of those DLLs is to answer the question "is this W8.1+?". The DLLs do nothing other than fake that. Maybe Python9 is similar.

Nah, unfortunately. This is a bunch of api-ms-win* DLLs here that were introduced in Win 8 or 8.1 and I don't think you can fake them to get any useful result. They contain a lot of system functions.
If someone manages to "fake" that, that would make a number of applications usable in Win 7, such as Python 3.9+, Vivado, KiCad, and many others...

Now I tried to build KiCad from source, that can be done on Windows with MSYS2, and would thus yield executables working on Win 7. This is documented there: https://dev-docs.kicad.org/en/build/windows-msys2/ , and I am used to building cross-platform apps using MSYS2, so no problem with that. So far, it won't build though, because KiCad 6 apparently requires the very latest version of wxWindows (3.1.5 IIRC), while MSYS still has wxWindows 3.0... so we'll have to wait till MSYS2 updates wxWindows. And then I'll keep you posted. If it does build, that will be a solution.

HI!

I've explained a working solution two or three pages earlier:
https://www.eevblog.com/forum/kicad/kicad-6-is-coming!/msg3897506/#msg3897506
"vcpkg/vs2019" is the officially way to build kicad 6.0.
If you enable the win7 compatibility patch you'll get a set of binaries which run fine under win7

[SiliconWizard]

Now I tried to build KiCad from source, that can be done on Windows with MSYS2, and would thus yield executables working on Win 7. This is documented there: https://dev-docs.kicad.org/en/build/windows-msys2/ , and I am used to building cross-platform apps using MSYS2, so no problem with that. So far, it won't build though, because KiCad 6 apparently requires the very latest version of wxWindows (3.1.5 IIRC), while MSYS still has wxWindows 3.0... so we'll have to wait till MSYS2 updates wxWindows. And then I'll keep you posted. If it does build, that will be a solution.

Wouldn't it be easier to use the Linux version of KiCad?
You'll have to dual-boot with windows 7 which is not ideal, but still saves you much time,
also because I expect monthly updates for KiCad V6.

Yeah no. If I'm going to use Windows, I don't want this hassle. Otherwise, I'd switch completely to Linux. It's likely going to happen, but not quite yet. I have a Linux headless box connected to the network, that I can access through ssh or VNC for GUI apps. But routing PCBs using VNC, no =)

Note that the freshly released KiCad 6.0 on some Linux distros is still a bit shaky. On the aforementioned box, there are issues with the schematic and PCB editor. We'll probably have to wait for a few more days until the first dependency quirks are ironed out.

I don't hugely mind having to build KiCad from source. That's something I've not done so far, but that I was considering doing anyway.

[SiliconWizard]

Building wxwidgets from source is not that difficult on Windows. Though I find it a bit ignorant that Kicad binaries don't work on Windows 7 out of the box. Most of the professional CAD software does because a lot of places still use Windows 7.

You're right about wxWidgets, but MSYS2 already had it as a package, and I want to limit the number of extra dependencies I have to build manually, especially when they would "clash" with the official packages. But yes, that would be an option!

As to Windows 7, yes. I can understand why many commercial vendors have now stopped supporting Windows 7, and now also build their apps so that they CAN'T even run on it (like Vivado), but for open-source software, it's a lot more questionable.

Ironically, Firefox, for instance, still perfectly works on Windows 7. And you'd think if the security of the OS was the real reason for stopping supporting it, a web browser would be much higher on the list than CAD software...

That said, for open-source software, distributed binaries are a bonus and they come with some limitations in a number of projects, because maintainers probably don't want the hassle of having to build for tens of different platforms, and test, even minimally, the artefacts. So if we have to build it from source - which certainly would be a major drawback for many people - I don't hugely mind. As long as building it is actually possible without having to go through multiple hoops, or having to use very specific tools and versions for everything.

[Smokey]

... A few pages back I complained that I couldn't get the 6.0 nightlies running in virtualbox mint linux 20.2. ...

Not sure if it's related, but I had similar issues with 6rc1 on an ubuntu virtualbox.  Just crashed a lot when I tried to open designs.  I figured out that you need to turn up all the graphics setting on the virtual machine.  Give it more graphics memory specifically.  Then it mostly worked.

[Smokey]

I'll further add to this ongoing tangent...
I'm sort of surprised that so many people are seeing dropping windows7 support as a show stopper.  Windows7 was good.  Accepted.  But windows 10 is also good.  And they give it away for free to all the windows7 users.  Really no great excuse to still being running windows7.  And I'm running one box with some old ass 2011 era hardware and windows10 works fine. 

[The Doktor]

Python 3.9, which itself doesn't support Windows 7 anymore.

I haven't looked into that yet, but my StreamDeck doesn't support Windows 7 in that it won't install, and if you extract the files it won't run because some DLLs are missing. The trick there is to shove fake DLLs in the install directory, and it runs. The only purpose of those DLLs is to answer the question "is this W8.1+?". The DLLs do nothing other than fake that. Maybe Python9 is similar.

Nah, unfortunately. This is a bunch of api-ms-win* DLLs here that were introduced in Win 8 or 8.1 and I don't think you can fake them to get any useful result. They contain a lot of system functions.
If someone manages to "fake" that, that would make a number of applications usable in Win 7, such as Python 3.9+, Vivado, KiCad, and many others...

Now I tried to build KiCad from source, that can be done on Windows with MSYS2, and would thus yield executables working on Win 7. This is documented there: https://dev-docs.kicad.org/en/build/windows-msys2/ , and I am used to building cross-platform apps using MSYS2, so no problem with that. So far, it won't build though, because KiCad 6 apparently requires the very latest version of wxWindows (3.1.5 IIRC), while MSYS still has wxWindows 3.0... so we'll have to wait till MSYS2 updates wxWindows. And then I'll keep you posted. If it does build, that will be a solution.

HI!

I've explained a working solution two or three pages earlier:
https://www.eevblog.com/forum/kicad/kicad-6-is-coming!/msg3897506/#msg3897506
"vcpkg/vs2019" is the officially way to build kicad 6.0.
If you enable the win7 compatibility patch you'll get a set of binaries which run fine under win7

Do you know of any place where I can download Win7 compatible binaries?

[PlainName]

But windows 10 is also good

Windows 10 is pretty much the same as Windows 7 under the bonnet so technically it is fine. The problem with it is that the user interface is shit, confusing and ignores even Microsoft's own guidelines. But it has to do that so it can be different and 'new'. There's also the forced updates, lack of quality control, the push online income streams, etc.

[SiliconWizard]

But windows 10 is also good

Windows 10 is pretty much the same as Windows 7 under the bonnet so technically it is fine. The problem with it is that the user interface is shit, confusing and ignores even Microsoft's own guidelines. But it has to do that so it can be different and 'new'. There's also the forced updates, lack of quality control, the push online income streams, etc.

Yep, that's it in a nutshell. The UI is atrocious, and it's not even pretty. The updates are bound to break things on a regular basis. The amount of telemetry is insane, and as some have shown, even when disabling everything that can be disabled, there's still some phoning home. The only sure way of stopping that is to cut the internet access, which, uh, is not really practical for everyday use.

If it was just a better 7, I would have long updated.

[Karel]

And as long as most people continue to use windows, microsoft gets away with it and nothing will change.

Only drug dealers and software companies call their customers "users".

[John B]

I started shifting over projects from version 5 to 6. Firstly I needed to add my custom symbols and footprints, but it's also worth noting there's quite a few changes to the standard library of symbols. Quite a few of the standard schematic symbols need to have the library reference updated, otherwise when you go into the PCB editor it will delete the footprint. I was partway through a project, so the schematic was complete but the PCB was half done. Probably my own fault there...

The 3D model references all need to be reselected too, otherwise nothing will show up in the 3D view of the PCB, but that seems to be purely cosmetic. The symbols in the schematic can be updated in bulk with only a few clicks but I can't see how to update all 3D models in one go.

Also, does anyone find the path finding ability when laying tracks almost broken? The PCB editor starts to lag badly after 1 or 2 bend in the tracks and never wants to lay a track through obviously large enough gaps unless its only 1 bend. There has obviously been significant changes in that regard compared to version 5.

[SiliconWizard]

I started shifting over projects from version 5 to 6. Firstly I needed to add my custom symbols and footprints, but it's also worth noting there's quite a few changes to the standard library of symbols. Quite a few of the standard schematic symbols need to have the library reference updated, otherwise when you go into the PCB editor it will delete the footprint. I was partway through a project, so the schematic was complete but the PCB was half done. Probably my own fault there...

The 3D model references all need to be reselected too, otherwise nothing will show up in the 3D view of the PCB, but that seems to be purely cosmetic. The symbols in the schematic can be updated in bulk with only a few clicks but I can't see how to update all 3D models in one go.

Also, does anyone find the path finding ability when laying tracks almost broken? The PCB editor starts to lag badly after 1 or 2 bend in the tracks and never wants to lay a track through obviously large enough gaps unless its only 1 bend. There has obviously been significant changes in that regard compared to version 5.

Thanks for the heads up. I've tried it on Linux, and only very little so far, and indeed opening any existing project made with v5 was a 'weird' experience. I can confirm the no 3D model and haven't investigated it so far, so thanks for the info.

All in all, there is a LOT of new features, new formats, new everything, and if you have to work on existing projects, at least professionally, I'd suggest waiting for a while...

[PlainName]

And as long as most people continue to use windows, microsoft gets away with it and nothing will change.

Only drug dealers and software companies call their customers "users".

As long as most people switch to Windows 10, you mean. Nothing wrong with Windows pre-8 that couldn't have been done without the screwup of Metro. But instead of holding out, like happened to force a rethink of Win8 and ME, everyone succumbs to the W10 giveaway and Microsoft's dirty tricks.

How's systemd going for you?

[ataradov]

How's systemd going for you?

I personally like systemd and it improved my life quite a bit. I had no issues with it, and creating services never been easier.

As far as migration goes, I just use my own libraries for everything to avoid issues like that. It is more work at the time, but pays off in long run.

[Cerebus]

Note that the freshly released KiCad 6.0 on some Linux distros is still a bit shaky.

I suspect that is "all platforms". The huge holes in the help files suggests that someone doesn't quite know what "finished and releasable" means. That kind of thinking about something so user visible doesn't bode well for what's going on behind the scenes.

e.g. The entire "Footprints and footprint libraries" section is all TODO.

7. Footprints and footprint libraries
7.1. Managing footprint libraries
NOTE
TODO: Write this section
7.2. Creating and editing footprints
NOTE
TODO: Write this section
7.2.1. Custom pad shapes
7.2.2. Footprint attributes
NOTE
Mention net ties here
7.2.3. Footprint wizards
For more information about creating new footprint wizards, see the Scripting section of the Advanced Topics chapter.

8. Advanced topics

It's not as if this is the only place, there are 29 occurrences of "TODO" in the pcbnew help file alone.

Personally I find this very embarrassing. I've been doing quite a lot of proselytising recently telling people who have previously tried and discarded KiCad things like "No, you should try KiCad again, they're getting better and better". Now I'm going to have to eat my words, "No, you were right, it is still amateur hour.".

[Smokey]

And as long as most people continue to use windows, microsoft gets away with it and nothing will change.

Only drug dealers and software companies call their customers "users".

As long as most people switch to Windows 10, you mean. Nothing wrong with Windows pre-8 that couldn't have been done without the screwup of Metro. But instead of holding out, like happened to force a rethink of Win8 and ME, everyone succumbs to the W10 giveaway and Microsoft's dirty tricks.
...

... Just think of the quality of horses we could have had if everyone had just "held out" on that whole automobile fad in 1909.   

I tend to agree with you that the metro UI is stupid, and I personally think Ubuntu Unity interface is worse, but it's crazy to say we should all be holding out on Ubuntu 10.10 from 2010 as some sort of protest.

[PlainName]

it's crazy to say we should all be holding out

That's the rational response to "as long as most people continue to use windows, microsoft gets away with it". The argument would be just which version we are discussing, and whether we let them or don't let them 'get away with it'.

However, much of the bad stuff that happens does so precisely because we 'let them get away with it'. If we don't nip it in the bud then whatever the bad thing is, it ratchets up to become the new normal and harder to stop. One upon a time a company called Phorm tried to force adverts on us by - get this - inspecting our Internet traffic for context! But they had to back down precisely because there was a big fuss about it. Fast forward to today and most people are quite content with the likes of Google and Facebook doing far worse than that, because they managed to slip the stuff by us in a way that only a few could object. We have let them get away with it.

It is crazy to think we should just sit back and let whoever do whatever they want without any hassle from us. No, if we are averse to their plans for global domination we should make that known and NOT let them get away with it. The problem is merely what qualifies for such action - how does the average pleb know that this is the cause to stand behind, which will (probably only temporarily) head off the later bad shit.

Nevertheless, we did almost do for Windows 10. Microsoft had to stiff Metro, at least notionally, and they still only managed to get it past us by giving the thing away free and actively stopping stuff working on Windows 7. We did indeed try to hold out, but weakness got the better of many of us, and here we are. (And I would like to take this opportunity to say "it is what it is".) 

[JohnG]

It's not as if this is the only place, there are 29 occurrences of "TODO" in the pcbnew help file alone.

Personally I find this very embarrassing. I've been doing quite a lot of proselytising recently telling people who have previously tried and discarded KiCad things like "No, you should try KiCad again, they're getting better and better". Now I'm going to have to eat my words, "No, you were right, it is still amateur hour.".

In Altium, they don't bother listing the TODOs, so you don't know they are missing until you get bitten by something. You are free to discover what is missing or undocumented by yourself. Maybe if you are lucky, you will find the solution on some online forum somewhere. Or, you can put in a ticket, and be told that "your problem is not actually a problem, according to our marketing". And if it's really a problem, you are SOL until the next version. Jest sayin'...

Kicad is not yet a replacement for Altium, but it sure seems on the right track. For my company, the key missing feature is database library support, so we continue to use Altium (which does the job for us, despite my complaining).

John

[djsb]

Instead of complaining about the help files, do something about it. I spent 3 years (2004-2007) translating the original French help files into English with the help of several volunteer translators. I started off using machine translation as I don't speak French.

[Cerebus]

Instead of complaining about the help files, do something about it. I spent 3 years (2004-2007) translating the original French help files into English with the help of several volunteer translators. I started off using machine translation as I don't speak French.

No, not having this. Answering a criticism of "Fundamental, basic, documentation is missing in what is supposed to be a release" with "You should write it" is an asinine response. Complete sections of basic help missing in a release? No, this is basic amateur hour - "Let's release this with no documentation at all on footprints and footprint tools and don't forget to not document zones or pads too". Literally everything that I've needed to look up in the help in the past few days has led to a "TODO". Would it be OK to "release" it without the footprint editor in place? No. Without tools to manipulate pads or zones? No. So why is it OK to "release" it without the documentation necessary to use that those?

4.4. Working with footprints
NOTE
TODO: Write this section - covers footprint properties, updating from library, etc.
4.5. Working with pads
NOTE
TODO: Write this section - covers pad properties
4.6. Working with zones
NOTE
TODO: Write this section

The truly stupid thing is that this documentation existed for 5.x. At least the 5.99 nightlies had the old helpfiles with basic information albeit out of date. So did 6.0 RC1. In 6.0 it's missing entirely. It would have been better to release 6.0 with the 5.x helpfiles in place with a caveat of "these are out of date" but someone decided to replace them after the release candidate stage with "new for 6.0" documentation that wasn't even half finished.

The issue here isn't that the documentation needs work to bring it up to date, the issue is that someone or some people made the decision to push 6.0 to "release" status with significant chunks of fundamental, crucial documentation missing entirely. It's telling that the response of a typical KiCad fanboy isn't "Oh, that's bad!", but "It's your fault becuase you didn't volunteer to fix it.". Volunteering to fix the documentation post facto wouldn't fix an incredibly poor, amateurish project management decision. In the commercial world a decision like that would get a project manager fired the next working day after the "release" got into the hands of customers with the letter from HR citing "reputational damage", "damage to the standing of the product in the marketplace", "costs to rectify", "loss of goodwill" and so on.

[JohnG]

In the commercial world a decision like that would get a project manager fired the next working day after the "release" got into the hands of customers with the letter from HR citing "reputational damage", "damage to the standing of the product in the marketplace", "costs to rectify", "loss of goodwill" and so on.

Yes, I suppose that this occasionally happens with commercial software.

What normally happens is that they circle the wagons around the executive who made the decision (and who got a bonus for early or on-time release), and maybe find some scapegoat engineer to fire if the heat gets bad enough...

John

[SiliconWizard]

Anyway, I managed building it on MSYS2. Actually, MSYS2 has both wxWidgets 3.0 and 3.1 packages, so it was a matter of forcing using 3.1 - needed some tweaking in one CMake file. After that, the build process went smoothly.

But was it all it would take? You bet your ass not. Kicad executables built this way run fine on Windows 7. But. Once you have passed the initial dialog asking you if you want to import your old settings (oh, I was happy when I got this dialog!), KiCad then greets you with a nice message "This operating system is not supported by KiCad and its dependencies. Any issues with KiCad on this system cannot be reported to the official bugtracker."

At this point, I thought, OK, annoying, but it's probably just a one-time warning and you can move on. You think? Nope! Once you click OK, the app just exits.
One can probably patch the source code to remove this, I'll have to see. But good lord, the KiCad team has gone to greath lengths to make sure Windows 7 users would suffer forever and not be able to use KiCad ever.

Yeah, I know some of you will say that it's perfectly fine, that Win 7 is dead, that... etc. Let's not go over this again, this has been done tens of times in other threads already. Point is, whereas I can understand it from a commercial vendor, I find it unacceptable for open-source software. Open-source software should never add anything to explicitely prevent people from building and/or running the software on any platform they see fit, under their own responsibility. I find this pretty bad. As I said earlier, even Firefox, which would be much more sensitive to potential security issues, has not done this.

So, I'll keep you posted, but I can say I do not like what they did here.

[voltsandjolts]

Point is, whereas I can understand it from a commercial vendor, I find it unacceptable for open-source software. Open-source software should never add anything to explicitely prevent people from building and/or running the software on any platform they see fit, under their own responsibility.

I see your point but I imagine their reasoning was something like 'moving to newer api's makes our life easier and we simply don't have the resources to support deprecated api's or systems'

I find this pretty bad. As I said earlier, even Firefox, which would be much more sensitive to potential security issues, has not done this.

I think your comparison is pretty bad.

Combined income of Mozilla Foundation and Mozilla Corporation
2015    $420 million       
2016    $520 million       
2017    $562 million       
2018    $436 million

I really don't think they took delight in screwing win7, they just felt they had to move on and put the resources they have to best use.

[SiliconWizard]

So, I looked up in the source code to locate this check, and got the bottom line. It's only, and entirely due to Python. From comments in the source code:

// Python 3.8 switched to Windows 8+ API, we do not support Windows 7 and will not
    // attempt to hack around it. A normal user will never get here because the Python DLL
    // is missing dependencies - and because it is not dynamically loaded, KiCad will not even
    // start without patching Python or its WinAPI dependency. This is just to create a nag dialog
    // for those who run patched Python and prevent them from submitting bug reports.

So it's nothing related to KiCad source itself, or even wxWidgets, but Python only. There's actually another app I use that requires Python, and the same problem - latest versions don't run anymore.

Now, under MSYS2, Python is at 3.9.7, and it works - so the MSYS2 Python package is built accordingly. This is why KiCad builds without an issue and has no problem launching... But still exits when it initializes Python. The reason is that I try running the built KiCad outside of MSYS2. One easy workaround is to run KiCad from within MSYS2 - actually, I noticed MSYS2 has a package for KiCad 6.0.0, so it should work there.

But I'd like to make it independent of MSYS2. And, I don't much care about Python scripting anyway in KiCad, so I'm currently re-building without this. Of course, it should also be possible to build Python 3.9 so that it runs on Win 7, and use that. Might look into that later.

I'll keep you posted... I guess a few people at least may be interested. And if not, at least you have the bottom line.