Re: [gentoo-dev] Re: My wishlist for EAPI 5

[Michał Górny]

On Thu, 21 Jun 2012 20:38:17 -0400
Richard Yao <ryao@g.o> wrote:

> On 06/21/2012 04:29 AM, Duncan wrote:
> > Richard Yao posted on Wed, 20 Jun 2012 16:50:33 -0400 as excerpted:
> >
> >> On 06/20/2012 04:35 PM, Ciaran McCreesh wrote:
> >>> On Wed, 20 Jun 2012 16:25:30 -0400 Richard Yao <ryao@g.o>
> >>> wrote:
> >
> >>>> POSIX Shell compliance
> >>>
> >>> So far as I know, every PM relies heavily upon bash anyway (and
> >>> can't easily be made not to), so even if developers would accept
> >>> having to rewrite all their eclasses, it still wouldn't remove
> >>> the dep.
> >>>
> >> Lets address POSIX compliance in the ebuilds first. Then we can
> >> deal with the package managers.
> >
> > Additionally, this is extremely unlikely because a number of
> > developers insist on bash, to the extent that it would likely split
> > gentoo in half if this were to be forced.  It wouldn't pass
> > council.  It's unlikely to even /get/ to council.
> >
> > Openrc could move to POSIX shell because its primary dev at the
> > time wanted it that way and it's only a single package.  However,
> > even then, doing it was controversial enough that said developer
> > ended up leaving gentoo in-part over that, tho he did continue to
> > develop openrc as a gentoo hosted project for quite some years.
> > Now you're talking trying to do it for /every/ (well, almost every)
> > package, thus touching every single gentoo dev.  It's just not
> > going to happen in even the medium term (say for argument APIs
> > 5-7ish), let alone be something practical enough to implement, soon
> > enough (even if everyone agreed on the general idea, they don't),
> > to be anything like conceivable for EAPI5.
> >
> > So just let that one be.  It's simply not worth tilting at that
> > windmill.
>
> Would you (or someone else) elaborate on the specific features of bash
> that people find attractive?

Local variables, reasonable behavior (like 'FOO=abc bar' where bar is
macro), arrays, [[ ]] tests (which are obviously faster than calling
external test program).

One more use: printing useful die messages (in POSIX sh there's no way
to do a backtrace).

-- 
Best regards,
Michał Górny