Re: [bug-gawk] Regression bug.

[arnold]

Hi.

Thank you for reporting a bug.

Without some way for me to reproduce whatever problem you're seeing,
I cannot proceed to find a solution.

Please send whatever instructions you think will help to reproduce the
problem.  Code and data would be even more helpful.

Thanks,

Arnold

Carlo Wood <address@hidden> wrote:

> Hi,
>
> after upgrading Ubuntu to Xenial, a script of me stopped working,
> which I traced back to a regression in gawk.
>
> I checked out the git repository, and did a git bisect:
>
> 31c6051694d3152e50eb037e20c4734c7321eac6 is the first bad commit
> commit 31c6051694d3152e50eb037e20c4734c7321eac6
> Author: Arnold D. Robbins <address@hidden>
> Date:   Sun Nov 16 19:54:57 2014 +0200
>
>     Add field reference changes. Currently breaks sortglos test.
>
>
> This makes sense because what happens is that some array
> starts to contain undefined / uninitialized values. So, I'm
> convinced that indeed this commit is buggy and caused the
> problem.
>
> Note that every version after this commit is bad, also current
> HEAD.
>
> Unfortunately, I cannot easily make a small test-case;
> I can however give instructions on how to reproduce the
> problem if you are interested in doing so.
>
> -- 
> Carlo Wood <address@hidden>