I'm encountering an issue with QTWebEngine, both with qutebrowser and with calibre, when building on default/linux/amd64/17.1/clang (applying the patch from https://bugs.gentoo.org/777147 to get a successful build), where QTWebEngine crashes on rendering certain pages (albeit not all of them). A simple "test" search on duckduckgo is enough to get this result. From discussions on #gentoo, it seems this also happens with falkon, although I haven't emerged it myself. This happens whether lto is enabled or not on qtwebengine (but I have not tried rebuilding the full qt stack without lto yet). I tried to get a backtrace through qutebrowser, but am not sure this was efficient, as gdb seems to treat QtWebEngine as a call to an external program. Happy to do other tests if needed. Reproducible: Always Steps to Reproduce: 1.launch qutebrowser 2.Search "test" in duckduckgo 3. Actual Results: Unable to load page. Error while opening https://duckduckgo.com?q=test&h… Renderer process crashed (status 5) Try again Expected Results: Display a list of results for a search for test.
Created attachment 864847 [details] Tentative backtrace This is a tentative backtrace obtained by running gdb -ex r --args /usr/bin/python3 -m qutebrowser --debug --temp-basedir |& tee gdb.txt
fwiw, for qutebrowser, you could switch to Qt6 -- I'd hope that it doesn't happen with qtwebengine:6 (need to use the live ebuild with USE=qt6 for this). Lot of web pages don't work with :5 either way due to too old chromium. Imagine calibre ebuild will switch to 6 whenever someone gives it attention too. Given qtwebengine:5 is essentially on life support, I wouldn't expect all that much from it.
I can confirm that it does work as expected on qutebrowser-9999 with QTWebEngine:6
