On Mon, 2023-08-14 at 17:48 +0200, priour...@gmail.com wrote: > From: benjamin priour <vultk...@gcc.gnu.org> > > Plenty useful, thanks David. I've adjusted some few things, especially > the artifacts of earlier versions I missed when building the commit. > > I didn't how to test for warnings within <memory>, I couldn't figure a > portable test. > I cannot pinpoint the line the warning is issued at in an inline DejaGNU > directive, > nor can I safely say the stack depth if I check a multiline-output (nor the > methods names) > > In the end, I found out an alternative, I am checking for the presence of > event "entry of 'main'". > Indeed, diagnostic_manager::finish_pruning comment's reads > If all we're left with is in one function, then filter function entry events. > The provided test case can only goes into main and std::* frames, so if > "entry of 'main'" exists, > it means we are also going into std::* frames. > > I've also adjusted the comment of prune_system_headers, analyzer.opt and > added an entry to invoker.texi. > > Successfully regstrapped off trunk > 54be338589ea93ad4ff53d22adde476a0582537b on x86_64-linux-gnu.
Thanks for the updated patch. This is ready to push to trunk. Dave