Conversation

Notices

1. Embed this notice
   Phel (phel@toot.cafe)'s status on Wednesday, 11-Sep-2024 15:11:27 JST Phel

   • Drew DeVault

   @drewdevault love it 😂😂 the U in GNU is for unhinged

   • Embed this notice
     Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Wednesday, 11-Sep-2024 15:11:23 JST Haelwenn /элвэн/ :triskell:

     in reply to

     • m0xEE
     • Drew DeVault
     • Gloopsies :fedora:
     @m0xee @drewdevault @gloopsies @phel Heh, earlier version of POSIX didn't have that one[1] and of course the reason is… GNU (and AT&T ksh *sigh*): https://www.austingroupbugs.net/view.php?id=1640
     
     1: https://pubs.opengroup.org/onlinepubs/9699919799/utilities/true.html
   • Embed this notice
     Drew DeVault (drewdevault@fosstodon.org)'s status on Wednesday, 11-Sep-2024 15:11:24 JST Drew DeVault

     in reply to

     • m0xEE
     • Gloopsies :fedora:

     @gloopsies @m0xee @phel per the standard true does not accept any arguments and that is important

   • Embed this notice
     m0xEE (m0xee@social.librem.one)'s status on Wednesday, 11-Sep-2024 15:11:24 JST m0xEE

     in reply to

     • Drew DeVault
     • Gloopsies :fedora:

     @drewdevault @gloopsies @phel
     Not entirely… hehe, pun intended🤭, true : https://pubs.opengroup.org/onlinepubs/9799919799/utilities/true.html
     "Whereas colon is required to accept, and do nothing with, any number of arguments, true is only required to accept, and discard, a first argument of "--". Passing any other argument(s) to true may cause its behavior to differ from that described in this standard"

   • Embed this notice
     Drew DeVault (drewdevault@fosstodon.org)'s status on Wednesday, 11-Sep-2024 15:11:25 JST Drew DeVault

     in reply to

     • m0xEE

     @m0xee @phel the entire purpose of true(1) is to unconditionally exit with status code zero

   • Embed this notice
     Gloopsies :fedora: (gloopsies@fosstodon.org)'s status on Wednesday, 11-Sep-2024 15:11:25 JST Gloopsies :fedora:

     in reply to

     • m0xEE
     • Drew DeVault

     @drewdevault @m0xee @phel

     Yes if you call it without arguments, but you called it with the --version flag, that changes what the program does, doesn't it? If you called `gcc --version 1.c` you wouldn't expect it to compile your program?

     翠星石 likes this.
   • Embed this notice
     m0xEE (m0xee@social.librem.one)'s status on Wednesday, 11-Sep-2024 15:11:26 JST m0xEE

     in reply to

     • Drew DeVault

     @phel @drewdevault
     Can you explain what's weird? I'm probably simplifying things here, but the program got passed an open file descriptor as stderr, it attempted writing into it, that call returned an error, the program aborted with a non-zero exit code, what is wrong with this?

   • Embed this notice
     Drew DeVault (drewdevault@fosstodon.org)'s status on Wednesday, 11-Sep-2024 16:04:33 JST Drew DeVault

     in reply to

     • Haelwenn /элвэн/ :triskell:
     • m0xEE
     • Gloopsies :fedora:

     @lanodan @gloopsies @m0xee @phel ultimately POSIX is subservient to implementations, not the other way around, for better or worse

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 12-Sep-2024 01:09:35 JST 翠星石

     in reply to

     • Haelwenn /элвэн/ :triskell:
     • m0xEE
     • Drew DeVault
     • Gloopsies :fedora:
     @drewdevault @lanodan @gloopsies @m0xee @phel GNU true's implementation is correct.
     
     It would be incorrect to return success if the write operation failed on --version.
     
     If you want it to return true - just don't request the version or help.
     
     If for some bizarre reason you want a version of true without flags that will tell you the license, you can just use GNU bash's true built-in.
     
     >One common use for true is replacing another program with it to shim or mock it so that it disregards the command line and always exits successfully
     In that usage, you won't be passing --version or --help to true, let alone piping to /dev/full or somewhere else that will fail to write, so that is a non-issue.