At Sunday 11 October 2009, Ralf Wildenhues <ralf.wildenh...@gmx.de>
wrote:
>
> Thanks for looking into this. I prefer not to paper over any bugs
> that we don't understand, though.
This is surely a good policy. I sent the patch only because I easily
came up with it (by five minutes of trial-and-error experimenting),
but it was only meant to be a temporary dirty fix, nothing more.
> If you are willing to look into this more, it's got to do with
> 'set -e'.
This is what I thought, but I couldn't easily find a minimal testcase
exposing this bug in heirloom-sh's dealing of `set -e'. Maybe I will
try again in the future.
> Note that while heirloom and Solaris sh are similar, they aren't
> identical, and I wouldn't want to invest work into fixing issues
> that are in the former only, since it's definitely not a normal
> development environment.
You're right about this, but please consider that the heirloom sh
offers a simple way to get immediate feedback about portability
errors w.r.t. Solaris sh, even for those contributors which don't have
an access to a Solaris or OpenSolaris system. Spurious errors
regarding only heirloom-sh can hamper this feedback. So I think it's
better if we ensure that the test scripts can work smootly with
heirloom-sh too, and not just to reach an only-therotically wider
portability, but to provide practical help to developers.
Regards,
Stefano
