On 2019/06/07 07:14, Rafael Sadowski wrote: > On Tue Jun 04, 2019 at 07:26:19PM +0100, Stuart Henderson wrote: > > On 2019/06/01 10:28, Rafael Sadowski wrote: > > > Update krita to 4.2.0. > > > > > > Krita 4.2 Release Notes: > > > https://krita.org/en/krita-4-2-release-notes/ > > > > > > To build it you have to deinstall krita 4.18 otherwise tests will fetch > > > the old one. > > > > > > Feedback is very welcome. All shared libs checked with > > > check_sym. Lightly tested on amd64. > > > > I hit this when testing build, sorry I didn't think to log it and it's > > well beyond my tmux scrollback. CMakeCache.txt gzipped and attached. > > Thanks for testing. It smells like our "normal" Cmake/Ninja build (re-)order > issue. > > I started from a clean setup with a fresh tree and no packages installed. > Built > fine. > > Is is a show stopper?
In a word, yes. If things fail in bulk they need looking into, so this is a lot of manual extra checking needed by people who are already putting a fair bit of time into running builds. Counting just the official bulk builds for amd64 and i386 plus exopi, that's probably about 10 builds a week, and I know there are other people who do larger or smaller bulks slightly less frequently. (In the case of ports which are depended on by other ports, a failure there blocking a bunch of other ports significantly slows down the build: check, fix broken port, restart, wait, come back and check again, fix another broken port, ....) If the breakage is too much, people doing non-essential builds are likely to just stop caring because they're too much work .. > > ons/pykrita/kritarunner/CMakeFiles/kritarunner.dir/__/plugin/pyqtpluginsettings.cpp.o.d > > -o plugins/extensions/pykrita/ > > kritarunner/CMakeFiles/kritarunner.dir/__/plugin/pyqtpluginsettings.cpp.o > > -c /usr/obj/ports/krita-4.2.0/krita-4.2.0/pl > > ugins/extensions/pykrita/plugin/pyqtpluginsettings.cpp > > /usr/obj/ports/krita-4.2.0/krita-4.2.0/plugins/extensions/pykrita/plugin/pyqtpluginsettings.cpp:19:10: > > fatal error: 'u > > i_manager.h' file not found > > #include "ui_manager.h" > > ^~~~~~~~~~~~~~ > > 1 error generated. > > ninja: build stopped: subcommand failed. > > > > ui_manager.h is auto-generated. As a hint: Usually moc_ and ui_ prefixed > files in Qt applications are generated files. The prefix comes from the > suffix. > > For example manager.ui: > > https://invent.kde.org/kde/krita/blob/master/plugins/extensions/pykrita/plugin/CMakeLists.txt > > Cheers, > > RS So if it's auto-generated, there ought to be something somewhere to say that pyqtpluginsettings.cpp depends on it. Is ki18n_wrap_ui meant to imply that (and if so is it broken?) Or is it just missing?
