Re: [Grep-devel] handling of non-BMP characters

[Jim Meyering]

On Sun, Dec 16, 2018 at 12:57 PM Bruno Haible <address@hidden> wrote:
...
> The test was running also on glibc, *BSD, and other systems.
>
> In fact, it was the *only* test that verifies that 'grep' handles
> beyond-BMP characters correctly.
>
> Now you have a test gap: If, by changes in glibc, in regex, in dfa,
> in the UTF-8 converters, or elsewhere beyond-BMP characters stop working
> on glibc or *BSD systems, no automated test will catch that.
>
> Therefore I would suggest to
>
>   - revive the test,
>
>   - rename it from 'surrogate-pair' to 'beyond-bmp' (to match what
>     it does, from a user perspective),
>
>   - change line 2 from
>       # Trigger a segfault-inducing bug with -i in grep-2.14 on Cygwin.
>     to
>       # Check the handling of characters outside the Unicode BMP.
>
>   - Add a comment
>       # Known failures: This test currently fails on Cygwin and AIX.

You're right. I'll restore the test and adjust as you suggest after the release.
Thanks,
Jim