On Thu, Feb 5, 2026 at 7:14 AM Nate Graham <[email protected]> wrote: > On 2/4/26 11:07 AM, Ben Cooksley wrote: > > Building Qt is quite the endeavour, it isn't for the faint of heart and > > in the case of WebEngine at least requires the VMLarge type on CI due to > > the memory requirements involved (16GB isn't enough for WebEngine), and > > also takes quite a long time. > > That would significantly complicate building and updating VM images for > > CI and make it significantly more resource intensive (not to mention the > > fact that we would probably then have to build all the things that > > depend on Qt, because we can't have the distribution install those > > anymore either). > > Doesn't Fedora turn on frame pointers by default these days?
> If we had a Fedora CI runner, we could run the LSAN code on only that > one. This would keep things fast for the other CI runners that don't > have frame pointers enabled by default, and we'd get LSAN results > without having to compile our own Qt. > > A Fedora CI runner would probably be useful anyway since it's a popular > KDE distro these days. > CI platforms have to essentially be used by everyone in order to provide the up to date dependencies. This is an artifact of the fact that various pieces of our software depend on the latest Frameworks, in addition to Plasma and Gear having interdependencies between each other, which often need to be latest as well. You therefore cannot have an image just for LSAN. With the quantity of dependencies we have for all the various applications in the KDE universe, changing CI platform is a huge undertaking, and platforms aren't chosen just because they're popular (usually it is for coverage of an area we don't have already, and realistically Fedora and SUSE are close enough to each other). Also, KDE software has a bad habit, especially in the Wayland space, of needing the absolute latest release version (this is why we use OpenSUSE Tumbleweed) so it would have to be Fedora Rawhide. > > Nate > Cheers, Ben
