Re: PATCH: subshell into a new process group

On Sat, May 9, 2026, at 1:30 PM, Mikael Magnusson wrote:
> Since you set the 0/0 min/max arguments in bintab, there's no reason
> to handle checking *args yourself (code should never be reached, you
> just happened to pick the exact same error message I think.

Very fair point, I was maybe just following a pattern, so happy to
revisit it.  I might have been thinking of laying groundwork for
supporting the non-zero arguments, since the setpgid naming does
hint at that.

> Why is the builtin called syssetpgid when you use setpgrp()?
>
> The manpage I have for these things says that setpgrp() is super
> obsolete and not supported, I guess this is the bikeshed part? I've
> never used these functions so I can't offer any particular wisdom but
> it seems like dragons be here territory to me.

Calling setpgrp() was just for consistency with the approach to naming
in the core, which does "#define setpgrp setpgid" where available (which
must be the overwhelming common case these days?) but then uses setpgrp,
either for max compatibility or just from inertia.  But the idea would
be, independent of the name chosen, that this patch should be safe to
add since it's just making the same call in the same form that is used
widely in the rest of the code base.  But yes, bikeshed indeed, I don't
know that implies the traditional old name is the right one to expose as
the command name.