Hey everyone,
Event: https://inkscape.org/cals/event/1/ Attending: Martin, Vaibhav, Ishaan, Philip, René, Tav Next: https://inkscape.org/cals/event/64/
CI builds with 26.04 from Ubuntu 24.04 and there are some breakages which will need attention. This was highlighted by a PPA failure with test failures. There were some ASAN issues, these might have effected our CI, but don't appear to be.
The test failures appear to be related to fonts. Martin talks about how capypdf works with fonts and how the system in the CI might be failing in it. We talk about the test failures for antialiasing and improving robustness.
Vaibhav has been working on backports. Wants to talk about a merge request for a crash ready for review.
Ishaan wanted to talk about the issue in Gtk4 clipboard and the missing functionality. We talk about the C++ bug fixing program and how much the contractor will have to fix bugs in upstream projects like Cairo, Gtk4 etc.
There's Gtk4 demos which we can test to see what can be fixed in Inkscape for 1.5.
Martin has continued working on the new rendering engine with a full separation of SPstyle using templating and a consistent API use they are now developing a way to allow the current rendering engine to print out the code for a running svg file so we can copy paste those outputs into test files to make sure we are getting the same output in the new engine.
Philip and Martin talk about the new engine's place, and how the discipline could help with using a GPU rendering engine in the future.
Pillip wanted to ask René about the CI improvements for pinning. The linux CI is not pinned and asks about the setup in use for the repository for ci building. There are details which this note taker does not understand, but it looks like a bunch of work are possible but would take time with GitLab.
Philip asks Tav about his color font merge request. There's an open merge request which can be fixed; any cairo version before 2027 would deadlock instead which we can patch in windows and macOS but not on Linux. Martin requests checks to avoid the new fix from causing the deadlock for users who might paste emoji into their document.
Philip finishes with the lib2geom test case issues; there's been a build up of test issues for various platforms. There's some slight differences between archetectures and Philip believes some of the features can't be fixed at all. Martin advises to contact previous lib2geom contributors at the same time as investigating the issues and seeing what is possible to fix and what tests will have to be disabled.
Tav is planning on a refactoring branch to cleanup the namespace use of Inkscape. Martin suggests that using namespace Inkscape is valid for a lot of code since it's part of Inkscape.
Ishaan reports that the 1.4.4 release in on schedule and thanks ltlnx (Willy) for their work on backporting. René would like to understand how the python interaction works for macOS for 1.4.4 but this might not be possible in the time.
Thank you everyone for coming to this week's meeting.