For a New PCB Software in Development

[GK]

Minus 20 IQ points for all those who posted a reply taking the OP seriously.

Minus 40 points for people saying others are stupid. Minus 60 points (even if that means going negative) for a 'no you can't do that' attitude.

You just forfeited 40 points without even realising it. Subtract another 40 points for the lack of self awareness.


 

Progress starts with really stupid ideas and unrealistic goals.

..... and an after hours team of developers whose experience and number must remain a mystery..............

 

[Mechatrommer]

the problem with joining KiCAD is, i dont like it at all. the workflow, the UI, esp the paradigm i guess. try to blend into existing code is one thing, try to modify it to suite your flavor is another thing. between doing major overhaul to existing one and start from scratch, yeah i'm with the OP, building from scratch (if you know about "coding" at all ). you may end up finishing either of them in the same amount of time, only the difference, with overhauling code you may end up with spaghetty+brocolli mix, but starting from scratch you may layout a uniform coding style from the beginning. but i havent joining any OSS project so i may be wrong. imho.

[HackedFridgeMagnet]

Just wild speculation.
It wouldn't be Nick Martin and a couple of friends would it.

Me and the team are seriously considering a new PCB software that reach all levels including "high-end professional market."

How much time before a release, this all depends on how things go as this is a project we do in our spare time. We have well designed code base and tons of code from other projects than we can use, so I can talk about 6 months to release a working demo.

[dannyf]

a new PCB software that reach all levels including "high-end professional market."

That means it is so comprised that it is of no use to anyone.

[poorchava]

There are different sorts of "high-end markets"...
-rf
-thermal
-high-speed
-mechanically demanding(think some exotic 20layer, metalcore, rigid-flex pcbs).
-autorouting

[EEVblog]

Just wild speculation.
It wouldn't be Nick Martin and a couple of friends would it.

Nope. Nick is working on an "Internet of Things" company, along with several other former faithful.
http://mesheven.com/
This is all that's out there on it it seems:

Mesheven Technologies is a stage-one startup focused on the Internet of Things (IoT) and the full range of related technologies. Mesheven’s hardware / software / cloud computing platforms will serve as the foundation for the creation of powerful device ecosystems that connect devices to the internet & make their data available for the organization of large-scale device “meshes”. This wealth of data will enable the creation of a powerful new generation of more intelligent systems that leverage the collection of data from multiple sources to make more intelligent ‘decisions’ about the world in which these devices operate. Mesheven’s solutions span the complete range of these technologies and provide the foundations on which other organizations can build their next generation of electronics products for this new world.

[mrflibble]

"hardware / software / cloud computing platforms"

really now. They forgot to include Big Data in the bullshit bingo. Well, at least they are leveraging stuff and spanning ranges so they got that part covered. Whatever happened to just saying WTF you're actually doing.  Ah well...

[dannyf]

There are different sorts of "high-end markets"...

If it is of use to everyone, it is of use to no one: because it takes a lot of intelligence and hard choices to understand your market, your customers, and make well-thought-out, intelligent compromises than to make a product for everyone yet no one.

[mrflibble]

the problem with joining KiCAD is, i dont like it at all. the workflow, the UI, esp the paradigm i guess. try to blend into existing code is one thing, try to modify it to suite your flavor is another thing.

That may be so, but the problem with yet another XYZ is that the new XYZ projects tend to repeat the mistakes from old XYZ projects.

"Ahahaah", says the OP, "no of course we will not repeat these mistakes because we are different". Of course you are different. Just like everyone else. I'm not saying that joining an existing project suddenly solves all things. There will be plenty of problems. But yet another one for the sake of yet another one doesn't make any sense. Unless this team has craploads of spare time and/or happens to be a huge team. A software package of this magnitude is no small feat. But whatever the choice, we'll see in 6+ months. Well, 1.5++ years after multiplication by pi to compensate for engineering estimates.

[nctnico]

As I understand I need to look deeper at Altium and do something better.

Correct, if you want any shot at the professional market.
But of course you would have already known this if you had any professional PCB design experience.
And like I said, Altium have been doing it for 30 years and still haven't got it right. So this is not a project you can expect to get anything really usable within a few years, let alone be in any way competitive.
I used to work at Altium, I know how much work goes into the software.
As another baseline, look at AutoTRAX (not the Protel one)
http://www.kov.com/
That took one guy more than a decade to get the this point, and no PCB professional would touch it with a 10 foot pole.

No wonder. The web page isn't very informative. Just some random screendumps showing what he can do with C#. What would be interesting is the number of layers, resolution, size of the design, flexibility in pad stacks, connection to a parts database, etc.

[Mechatrommer]

Unless this team has craploads of spare time and/or happens to be a huge team. A software package of this magnitude is no small feat. But whatever the choice, we'll see in 6+ months. Well, 1.5++ years after multiplication by pi to compensate for engineering estimates.

i believe kicad team is not small but they are doing it very slow and the result is not that magnificent imho. the problem is not the development itself, but "the specification" from the beginning which a corporate usually do it in round table meeting. for example how the GUI work, how mouse is used to minimized clicks and movements etc. i even having trouble export schematics to pcb hell i hate to read guide.pdf they keep insisting my document folder is lock or whatever. meh i hate to talk more it just reveal more of my stupidity.

otoh take diptrace as another example, it is better than kicad at several magnitude imho, but still far behind altium. somehow diptrace was made from a well defined (or better) spec, it is on going improvement, similar to any other SW development, free or not free. i believe whats driving them the developers is the fact that it is a paid version, but the price is what makes me stay away from diptrace. the most plausible OSS on the planet is linux but that also i can give one or two weaknesses which is strong enough for me to stay away. kicad is far from diptrace and even farther from altium...

a better product need to be built right from the ground zero and kicad is not it imho. they can continue to put their ego down while continuing this project and we can wait where they are heading. thats my long story short.

[andyturk]

The last thing I want is that some random hippie fart could mess up my library by storing something in some fscking cloud.

What the heck is a "random hippie fart"? Yowza!

Is it a result of granola, birkenstocks and/or hash oil?

[mikeselectricstuff]

the problem is not the development itself, but "the specification" from the beginning

Exactly.
Like doing a PCB layout - get the placement right and the routing is easy.
It's all about the design and architecture. Get that really right and the coding is near-trivial.

[AndyC_772]

I'm not sure I'd go that far, but the reverse is certainly true: get the underlying architecture wrong and all the subsequent effort in the world won't give a usable product.

I have no idea how a really fast, efficient, usable 'push & shove' routing algorithm works, and I certainly wouldn't have the first clue how to go about implementing one. But I know they exist, and that having a really good one is a bare minimum requirement for any serious layout package. It's what the board designer will be using 99% of the time once the board is placed and the DRC rules are set up, so it's got to be absolutely right. The underlying database structure has to be designed around that.

If it's not as fast, capable and reliable as the one in Allegro then I'm not interested, sorry.

[zapta]

Hell, why even use that when you have rsync + ssh? No need for that fancy pants torrent stuff. Add rsnapshot to it, and you're all set. Cloud, shmoud indeed.

This is called private cloud http://en.wikipedia.org/wiki/Cloud_computing#Private_cloud

[mrflibble]

Well, it was called rsync+ssh with a dash of rsnapshot well before the twat department decided to slap a Cloud Computing sticker on anything even remotely resembling 2 networked nodes with <whatever> running on it. So I shall keep calling it rsync+ssh with a dash of rsnapshot.

I would also do some of that Elastic Computing, but I'm afraid the rubber band might snap. On my wallet. So semi-local non-elastic computing it is. Although EC isn't as bad in the hype department as the aforementioned cloud shmoud.

Aaaanyway, on the PCB software. The trick as with all software would indeed be to 1) get the architecture right, and then possibly even more importantly  2) have a benevolent dictator on the team that fiercely guards this architecture and oversees implementation. Open source software tends to suffer from the Too Many Master Chefs In Kitchen affliction far more often than corporate software. And they both from Not Invented Here Syndrome in about the same amounts I'd guess.

So one smart guy with a vision, one asshole with same vision and several moderately compliant yet capable programmers == win.

[Monkeh]

Hell, why even use that when you have rsync + ssh? No need for that fancy pants torrent stuff. Add rsnapshot to it, and you're all set. Cloud, shmoud indeed.

This is called private cloud http://en.wikipedia.org/wiki/Cloud_computing#Private_cloud

No, that's just a label they've tacked onto a practice which existed long before the 'cloud'.