[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [CFT] Shell functionization patch (if you don't know what that means
From: |
Ralf Wildenhues |
Subject: |
Re: [CFT] Shell functionization patch (if you don't know what that means, it's faster Autoconf and leaner configure scripts) |
Date: |
Sat, 18 Oct 2008 08:46:08 +0200 |
User-agent: |
Mutt/1.5.18 (2008-05-17) |
* Paolo Bonzini wrote on Fri, Oct 17, 2008 at 09:18:30AM CEST:
>
> > [...] you will want to put something like
> > $as_debug
> >
> > at the start of each function (either being empty, or 'set -x'), for
> > ksh.
>
> Do you want to put that in AS_REQUIRE_SHELL_FN? It's fine by me.
I'd love to. However:
- either this is a totally undocumented interface, in which case the
desire to have it should come from maintenance need,
- or it aims to be documented eventually; in that case, there is the
need to deal with potential breakage resulting from
set -x
$command >$log 2>&1
$evaluate $log
compare the hoops Autotest goes through for this.
> > So, whenever you expand (or eval) arbitrary text into a shell function,
> > there is a potential problem.
>
> But it is not arbitrary text. See my message in the other thread.
>
> > I'm not quite sure how prevalent systems with this zsh are (but probably
> > still too common to completely ignore them). Even when we can ignore
> > them, the script header should be sure to error out for such systems
> > (because the corruption is silent, and hard to detect).
>
> ... and in fact IIRC this shell will be ignored by the better-shell loop
> because it fails one of the tests; I think it is as_func_failure:
Indeed.
Thanks,
Ralf