> On 4 Nov 2022, at 12:26, Adhemerval Zanella Netto via Libc-alpha wrote: > > > > On 04/11/22 09:10, Florian Weimer wrote: >> * Adhemerval Zanella Netto: >> >>> On 04/11/22 04:15, Florian Weimer via Libc-alpha wrote: >>>> Some sources merely include without -D_GNU_SOURCE and expect >>>> declarations for posix_spawn_file_actions_addchdir_np to be available. >>>> >>>> Tested on x86_64-linux-gnu. >>>> >>> >>> LGTM, although you reference posix_spawn_file_actions_addchdir_np on commit >>> message, but you export posix_spawn_file_actions_addtcsetpgrp_np. >> >> Sorry, there are multiple functions that are now available by default. >> I encountered the problem just with posix_spawn_file_actions_addchdir_np, >> so I still think the commit message is correct. > > I would advise cite all the function affected then, because with only > posix_spawn_file_actions_addchdir_np on commit message it might seems that > the patch is exporting more than intended. > Agreed. I'm a little bit uneasy given this ends up masking some problems and then we're back to having to only fix them on musl systems, but if this is the only function not exposed and the logic already applies to the others, I guess it's fine. (In particular, there's various functions from glibc where I have to go around and spot missing _GNU_SOURCE right now for Clang 16/modern C porting, but if we pursue more changes like this, we end up having to do more work to find issues on other platforms like musl instead of getting it for free.) thanks, sam