[Toybox] TOYFLAG_ARGFAIL

[Rob Landley]

On 3/14/19 12:42 AM, enh via Toybox wrote:
> according to the coreutils docs, chroot/env/nice/nohup all have the
> same "125 for command errors" behavior. that makes sense to me, and my
> OCD wants to send you the patch, but i don't know whether that
> conflicts with your dislike of adding stuff no-one's asked for.

It's fine. I wouldn't have added ARGFAIL without a user, but once it's added and
has a couple users, filling out more while we're thinking about it is just
following through. It was created for real users, and filling out similar cases
"while we're there, for consistency" is entirely reasonable.

> timeout and grep are special in that we know people do check their
> exit status, so i'm not sure whether we have precedent yet for this
> behavior in general.

We already do it in some places, we already paid the cost of the infrastructure
to do it (mostly conceptual cost in this case, the implementation was trivial
but it's more buttons on the Tardis console and these are _not_ telepathically
labeled so the Right People know what they do at a glance without ever having
seen that layout before)...

Doing it everywhere we need to makes more sense than doing it in just some
places. Consistency is a form of simplicity. :)

> (as best i can tell, there aren't actually that many more such cases.
> su does 125, runcon probably ought to but doesn't. cmp/diff/sort are
> like grep with 2. ls and expr are both [but differently] weird. that
> seems to be it?)

Strike while the iron is hot and all that. If you already did the analysis I'll
happily apply the patch.

Rob