Re: ZILE 2.3.10 (Autoconf 2.64 finds deadly DJGPP Bash bug)

[Rugxulo]

Hi,

On Sat, Aug 22, 2009 at 2:44 AM, Ralf Wildenhues<address@hidden> wrote:
>
> * Eric Blake wrote on Fri, Aug 21, 2009 at 01:51:29AM CEST:
>> Or does it require a more complex function body, to show the difference
>> where the subshell avoids the crash?  At any rate, we'll need to
>> characterize exactly what your shell's bug is.
>
> Yes.  Rugxulo or Reuben, can you please post a link to the package
> that's failing?  Thanks.

http://ftp.gnu.org/gnu/zile/zile-2.3.10.tar.gz

>>> However, the whole point of using 'return $ac_retval' without a subshell
>>> is to avoid forking, and catering to bash 2.04 brokenness (if it is indeed
>>> a bash bug fixed for 2.05a, as I suspect) is a step backwards.
>>
>>
>> You really think that this will slow everything down? Have you tested
>> it on your machine? (hint hint)    :-)
>>
>> Yes I have tested it on cygwin, and yes it made a difference, which is why
>> we committed the patch in the first place.  Seriously - the cost of a
>> subshell is VERY measurable, and doing anything to avoid a subshell will
>> have NOTICEABLE differences in the running time of a configure script.

Cygwin doesn't count as it's slow anyways, already emulating a lot of
stuff (e.g. fork) because Windows isn't POSIX friendly. Please
consider trying again on a real *nix (Linux, presumably), and tell me
if it takes more than a few secs extra, if even.

> We could use as_fn_set_status everywhere, and only define that to
> use a subshell if we detect 2.05a.  That shouldn't penalize users of
> non-broken shells so much.  (Of course, this is pretty vague given
> that we don't know what exactly the bug is.)

Both DJGPP ports of Bash 2.04 and 2.05b have this issue.

GNU bash, version 2.04.7(1)-release (i386-pc-msdosdjgpp)
GNU bash, version 2.05b.0(1)-release (i386-pc-msdosdjgpp)

Yes, it is vague, but I don't know what would be causing this either.