Start with 5.1.7 or 5.99?

Finally have time to start doing some PCB design work.

Should I start with the 5.99 nightly or 5.1.7?

I’m quite familiar with PCB layout tools and concepts.

5.1.6 is the current release version though some would say the bug fixes and apparent stability of 5.1.7 make it a better choice.

5.99 is only for testing and you are willing to accept you could lose all your work.

Even if you accept data loss, the file formats are in flux. It’s not even guaranteed that you can continue your design from one nightly build to another. So it’s really only for testing. That said, the files can be modified with a text editor, so if you do something for hobby and at the same time want to help a great Open Source project by testing and reporting bugs, you can ask for help here if you want to use 5.99 and get stuck.

Thank you. I get the concept of nightly builds, 5.1.x incompatibility, etc.

I was wondering if 5.99 / nightly were sufficiently stable to do work? I don’t mind a crash a week if it helps move the project along.

I quote myself:

How about a crash a day or hour? So that you can’t continue the workflow in some direction?

FWIW, I’ve been using 5.99 from the start. But electronics are just a hobby for me, and I save often…

There is no such thing as 5.1.7 yet.
There is a frequent Testing release that is a pre-5.1.7 snapshot. Usually but not always, these are safe to use and often better than 5.1.6 as several bugs have been fixed.
I put up threads from time to time identifying releases that I have found to be good.
File formats do not change in this fork, so your project can be shared with 5.1.6 users

5.1.6 was the last stable version.

FWIW, I’ve been using 5.99 nightlies basically from the beginning and I did not experience anything catastrophic. There were bugs/crashes that would stop you from doing something meaningful, but most were fixed literally within hours and usually made it to the next nightly.

But then I’m a hobbyist as well. Saving often - as already mentioned - is highly recommended. Maybe even keeping a couple of days of backup.

If we just take the original question and try to aswer it shortly, the answer is easy and clear: use 5.1, possible testing builds post-5.1.6. Otherwise the original poster should now have enough information and opinions to make the decision. If not, we need to know more about the situation.

Highly recommended even for 5.1.6, which does contain some rare but nasty bugs

I accidentally installed 5.9 on one of my systems. It wouldn’t work with my library. I don’t know if it was me or the program. Finally get it reverted back to 5.1.6. I hope that fixes it. I’m not exactly a genius on library creation in the first place.

1.5+ months later, is it ready enough now?

No, if you’re coming from v5.1.x

If you’re coming from ground zero: Maybe.

Summary:
I would not recommend any serious production work on 5.99. I just went through evaluating if I could just work around the bugs as I’ve been with KiCad since pre-v4 and can deal with a lot, but I would say it’s possible but not worth it for someone who is on a time budget. If you’re doing only hobby work, that would probably be just fine.

Schematic Editor:
Similar enough to v5.1.x. There are definitely nice upgrades and things that will save time…but then there are enough usability bugs to slow you down. Overall, the improved workflow is not a time saver due to dealing with having to workaround bugs.

Library:
If you have your own library of parts, doing anything to symbols or footprints in v5.99 will prevent the library from working in 5.1.x. And the forward-compatibility is likely, though not guaranteed.

Layout:
Had the program crash a few times when updating the PCB from the schematic. Couldn’t tell why, but could work around it if I was careful to click the buttons in the same order every time. Makes updating the PCB a guessing-game. Routing tracks was the standard push-n-shove, but similar to schematic, there are enough weird bugs that makes using it not a time-saver over v5.1.x.

Plugins:
I could not get teardrops or layout replication to work. I stopped trying plugins at that point since 2 of 2 plugins didn’t work.

Other Notes:
3D viewer seemed to work fine. It output gerbers that looked usable, but I did not mfr the board to tell. If I were to do a full design start-to-finish, I would expect about a crash-an-hour on average.

Conclusion:
I’m still very excited about KiCad v6, and the work that’s been done on it so far is incredible. That being said, it is understandably in the pre-release state as the feature-freeze just happened. If you’re looking to schedule your time with a known workflow, go for v5.1.x. If a program crashing on you multiple times a day bothers you, go for v5.1.x. If you’re ready for the latest-and-greatest and are more hobby-level, 5.99 is pretty cool.

3 Likes

I am not what you would call a power user of KiCad (although I design power). I loaded 5.99 when I was not working and later I got some consulting work. At that point I did not want to try to revert to a stable release build. While some of the 5.99 builds have been problematic, the one I have been using has played well with one crash over about 2 months of light to moderate use. When a given build was problematic I updated it and sometimes reported a bug. When I found one which was good, I stuck with it and stopped updating at kicad-r17012.9a4f92dfd9-x86_64-lite. This was an easy and practical solution for my situation. The point is that some of the nightly builds seem to be much better than others.

In 5.99 I like the highlighting of selected areas of a schematic diagram. I have ordered and received boards which appear to be good.

1 Like

People always say this, but please point to the last bug that actually resulted in these consequences?

Several months ago, the Testing nightly would crash every time I moved a footprint - this was a ratsnest bug. It was fixed a few days later.
I was move -save - move - crash - restart

These things happen from time to time

Are these usability bugs reported? The reason I ask is because we’re not aware of some big backlog of usability bugs that need fixing in eeschema

And the forward-compatibility is likely, though not guaranteed.

If by forward-compatibility you mean creating libraries in 5.1 and then opening them in 5.99/6.0, yes that is guaranteed.

I could not get teardrops or layout replication to work. I stopped trying plugins at that point since 2 of 2 plugins didn’t work.

There have been enough changes to the internals that most Python plugins will need to be updated for V6, and I imagine this won’t happen for many until after there is a release candidate

If I were to do a full design start-to-finish, I would expect about a crash-an-hour on average.

As far as I’m aware, there is nothing in KiCad that causes a scheduled crash every hour. Instead, there are specific actions that are currently broken and need to be fixed. please report these if you do not see them reported already, do not assume that we know about them.

1 Like

For starters, unless there’s a certain feature in the “5.99” that is mandatory for someone, one should start with 5.1.x release.
It’s complete enough for serious work and stable for everyday work, and at the time 6.0 will become mature, it should be relatively easy to upgrade.
The other way round, there may be “dragons ahead” now on the 5.99 route and the “downgrade” of the project will be hard as it’s not supported.

@nikolx - thank you for that AWESOME response.

I’ve been thinking about this and I’m going to use the 5.99 nightlies.

I don’t have many ways to give back to the project and possibly running into a few issues (as long as the board turns out fine and it isn’t too bumpy), is an incredibly low price to pay for the wonderful and selfless work being done by the developers and community.

I figure the more people that party on 5.99, the faster it will stabilize and the better the releases will be.

Is there a way to look at regression test results for the nightly build so that one can be smart about upgrading?

All the best.

2 Likes