mbswidth "failure" on Solaris

[Akim Demaille]

Hi!

Bison uses mbswidth to track locations in grammar files.  We recently
strengthened the test suite on this regard.  It is reported to work
well on GNU/Linux and MacOS, but Kiyoshi reported a failure on Solaris
with UTF-8 (http://lists.gnu.org/archive/html/bug-bison/2019-04/msg00020.html):

> 127. diagnostics.at:130: testing Tabulations and multibyte characters ...
> ./diagnostics.at:130: perl -p -e 's{</?\w+>}{}g' <experr.orig >experr
> ./diagnostics.at:130: LC_ALL=en_US.UTF-8 bison -fcaret -Wall input.y
> --- experr      2019-04-30 07:15:36.726301382 +0000
> +++ /tmp/bison-3.3.90/tests/testsuite.dir/at-groups/127/stderr  2019-04-30 
> 07:15:36.759788751 +0000
> @@ -10,9 +10,9 @@
> input.y:14.4-17: warning: empty rule without %empty [-Wempty-rule]
>     14 | d: {éééééééééééé}
>        |    ^~~~~~~~~~~~~~
> -input.y:15.4-17: warning: empty rule without %empty [-Wempty-rule]
> +input.y:15.4-20: warning: empty rule without %empty [-Wempty-rule]
>     15 | e: {∇⃗×𝐸⃗ = -∂𝐵⃗/∂t}
> -      |    ^~~~~~~~~~~~~~
> +      |    ^~~~~~~~~~~~~~~~~
> input.y:16.4-17: warning: empty rule without %empty [-Wempty-rule]
>     16 | f: {   42      }
>        |    ^~~~~~~~~~~~~~

The error (three more columns than expected) seems to indicate something
related to the combining arrow.  Maybe I'm expecting in the test suite
something which is not guaranteed to work universally?

Kiyoshi reported what seems to be proper support of UTF-8, but maybe
the problem is in the Unicode version supported by the platform?

> $ LANG=en_US.UTF-8
> $ export LANG
> $ LC_ALL=en_US.UTF-8 locale
> LANG=en_US.UTF-8
> LC_CTYPE="en_US.UTF-8"
> LC_NUMERIC="en_US.UTF-8"
> LC_TIME="en_US.UTF-8"
> LC_COLLATE="en_US.UTF-8"
> LC_MONETARY="en_US.UTF-8"
> LC_MESSAGES="en_US.UTF-8"
> LC_ALL=en_US.UTF


Thanks!