Up until the 2 most recent commits to the master branch my inhouse footprint library makes kicad go belly up.
Any take on what might have changed?
I also noticed that the kicad file format appears to have changed.
In some cases kicad wants to save in the new(updated?/amended?) file format.
Wonders and more wonders.
Can you post the info on what version you are currently using please . . .
(Help>About KiCAD>Copy Version Info . . . then Paste here)
I also noticed that the kicad file format appears to have changed. In some cases kicad wants to save in the new file format.
The master branch is always the current development version. This is actually the v7.99 version. Because this version is under active development new features are added every minute. And for new features the file-format must be enhanced with more tags. So a change in the file format must be expected for the master branch version.
This will happen until late autumn, than a file format freeze is announced and from than on the file format should be fixed.
If you want now a stable fileformat you need to work with the stable kicad version - this is currently v7.0.x.
Apologies, should have thought of that!
The ‘dirty’ would mean I set EGL to ON in the CMakeList.txt file
Application: KiCad x86_64 on x86_64
Version: 7.99.0-1601-g3a0f8214fa-dirty, release build
Libraries:
wxWidgets 3.2.1
FreeType 2.11.1
HarfBuzz 2.7.4
FontConfig 2.13.1
libcurl/7.81.0 GnuTLS/3.7.3 zlib/1.2.11 brotli/1.0.9 zstd/1.4.8 libidn2/2.3.2 libpsl/0.21.0 (+libidn2/2.3.2) libssh/0.9.6/openssl/zlib nghttp2/1.43.0 librtmp/2.3 OpenLDAP/2.5.14
Platform: Ubuntu 22.04.2 LTS, 64 bit, Little endian, wxGTK, xubuntu, x11
Build Info:
Date: Jun 23 2023 17:04:40
wxWidgets: 3.2.1 (wchar_t,wx containers) GTK+ 3.24
Boost: 1.74.0
OCC: 7.5.2
Curl: 7.81.0
ngspice: 31
Compiler: GCC 12.1.0 with C++ ABI 1017
Build settings:
KICAD_USE_EGL=ON
I quite realize that. I do a daily pull followed by a recompile.
The actual point being here is that up until the last 2 (days) commits all has been well.
Totally agree, however, this still raises the issue as to how to update to exactly that what has been changed.
I could wade through the format defninitions and compare with my foorprint lib (261 items). It would be kind of tedious to find the bugger.
The issue of change appears to be related to ‘footprint reference properties’ orientation manifesting in pcbnew on loading the pcb.
No matter how much I set them in the intended orientation they keep their own mind.
e.g. a 90⁰ orientation ends up as hozicontal 
Even on 7.99, a change should cause KiCad to reject your library nicely, not just crash.
I kind of agree on that.
I would think that raw and very unfinished code is worked on by the individual developers in their own local branched source tree of the master branch, and when the developer feels the changes he makes have reached a level of functionality that is worth trying out for unknown bugs by adventurous users, it is pushed to the public KiCad master branch on gitlab to end up in the nightlies.
And that would then imply that code in the public source tree is there in order to actually get the feedback from the users as soon as possible, to be able to fix issues immediately while the particular parts of the code are still fresh in the memory of the developer.
Which would then mean that users are invited to report bugs on the master branch / nightlies as soon as possible rather than wait to the final moment before the release of the stable 8.0…?
The nightlies are currently very unstable due to some big architecture changes going in. Stability should be trending back upwards over the coming week.
Just out of curiosity, would you mind describing these architecture changes in a couple words or sentences?
It’s a bunch of separate unrelated things, it’s probably easiest to satisfy your curiosity by looking at the commit log: Commits · master · KiCad / KiCad Source Code / kicad · GitLab
I think that Master gets to Alpha stage when the file format freeze happens, in the Autumn.
It’s a sandbox at the moment
In the meantime is there a place where to find some file format deltas?
It also would be very helpful if kicad reports the offending line instead of just go bellly up.
Apart from that. Issue still present as of toadays pull with last commit b8e95448
Reads like a weather forecast!
The big architecture changes were the use of SCH_COMMIT in the schematic editor (which follows the PCB Editor use of BOARD_COMMIT), and the introduction of PCB_FIELD in the PCB Editor (which follows the schematic editor’s SCH_FIELD).
[Edit: oh, and also the use of std::any instead of void* in the TOOL_EVENT system.]
By some ‘magic’ library loads again. Thanks.
However, component texts in pcbnew are still at random angels.
Thanks for the info. That’s exactly what happens.
Curious however is that this issue did not manifest up until about a week ago.
See my post above about large changes going in to hopefully address your curiosity.
