Summary: | dev-libs/boehm-gc-8.2.2[threads]: fails to build (error: too few arguments to function ‘pthread_setname_np’) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Cănărău Constantin <canarauc> |
Component: | Current packages | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
Status: | RESOLVED NEEDINFO | ||
Severity: | normal | CC: | patriellabork |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
build.log
config.log |
Description
Cănărău Constantin
2022-08-29 08:29:50 UTC
I feel like there's something more to this given I can't reproduce it with the same USE flags. # ifdef HAVE_PTHREAD_SETNAME_NP_WITHOUT_TID /* iOS, OS X */ (void)pthread_setname_np(name_buf); Why did it take this path for you in /var/tmp/portage/dev-libs/boehm-gc-8.2.2/work/gc-8.2.2/extra/../pthread_support.c? My configure gives: >checking for pthread_setname_np... yes (with tid) Yours gives: >checking for pthread_setname_np... yes (w/o tid) Please share your config.log. Have had the same error. Removing "-w" from CFLAGS fixed it. Created attachment 801970 [details]
config.log
@Sam James config.log is attached. @Tino Mueller Yes, filtering -w does solve the issue. I considered -w just cosmetic, so I didn't remove it form plain flags. Thank you all for support. (In reply to Cănărău Constantin from comment #5) > @Sam James > config.log is attached. > > @Tino Mueller > Yes, filtering -w does solve the issue. > > I considered -w just cosmetic, so I didn't remove it form plain flags. > Thank you all for support. I could add a filter to the ebuild but the problem is, other packages might silently be misconfigured with -w because configure tests often rely on warnings. What I recommend you do is use -w in a hook that applies it only after src_configure, to not affect configure tests. I have a cheesy version of this for -Werror=implicit-function-declaration (which also breaks configure often). -w is there from a time with (much) older portage when I also used -fail-clean FEATURES and size matters. Right now, this is not the case. I will remove -w from C(*)FLAGS. Also, do not know if filter in the ebuild is really necessary. I waited few days before to fill the bug, due to my aggressive cflags and nobody complain or filled a bug. But I am not a Gentoo developer. Thank you for your support. |