Re: getopt() not exposing __optpos - shell needs it

[Rich Felker <dalias@libc.org>]

On Tue, Aug 29, 2017 at 06:47:24PM +0200, Denys Vlasenko wrote:
> On Tue, Aug 29, 2017 at 3:07 PM, Rich Felker <dalias@libc.org> wrote:
> >> > Maybe I'm missing what you're trying to say, but all the state is
> >> > clobbered; I don't see how optarg is a problem specifically. You can
> >> > clear or set it to a sentinel value before the relevant call if you're
> >> > trying to determine if the call set it. Across other calls (not the
> >> > one for the current option) I don't see why it matters at all what
> >> > happens to it.
> >>
> >> Yes, this can be done.
> >>
> >> It gets increasigly ugly, though.
> >>
> >> With these amounts of massaging around libc API design breakage,
> >
> > Yes the getopt API is horribly broken. It's all global state, with a
> > tiny portion of that state internal/inaccessible. It doesn't follow
> > that the solution is adding new extensions every time an application
> > hits an obstacle from the brokenness. The right direction for fixing
> > it on the libc side would be introduction (with consensus across
> > important implementations) of a getopt_r API or similar with no
> > global/internal state.
> 
> I don't understand why you are opposed to exposing __optpos.
> It does not even require any coding. Not a single insn needs
> to be added.

New public interfaces are a lot more expensive than new code. The
latter can be changed or removed; the former can't.

Over the past 6+ years, just about the only other party asking musl to
add newly-invented nonstandard interfaces without existing precedent
was gnulib, and after a lot of discussion that made sense because they
were already doing the equivalent on other libcs through hideous
poking at internals that were never meant to be exposed, and because
the new interfaces fixed compatibility with tens (or more) of packages
using gnulib that weren't even aware of its dependencies on poking at
internals. OTOH you've requested this kind of thing for busybox hush
twice over just a month or so, and in this case the (getopt) the same
can clearly be achieved portably through the existing interfaces so
there's not even a need.

Rich