Re: texinfo-6.7.90 pretest

[Gavin Smith]

On Sat, Feb 27, 2021 at 12:11:05PM +0200, Eli Zaretskii wrote:
> 3. Finally, an error that stopped the build:
> 
>      if TEXINFO_DEV_SOURCE=1 ; export TEXINFO_DEV_SOURCE ;  top_srcdir=".." ; 
> export top_srcdir ;  top_builddir=".." ; export top_builddir ; 
> /d/usr/Perl/bin/perl ../tp/texi2any   -I . \
>       -o texinfo.info `test -f 'texinfo.texi' || echo './'`texinfo.texi; \
>      then \
>        rc=0; \
>      else \
>        rc=$?; \
>        $restore $backupdir/* `echo "./texinfo.info" | sed 's|[^/]*$||'`; \
>      fi; \
>      rm -rf $backupdir; exit $rc
>      Your vendor has not defined POSIX macro LC_MESSAGES, used at 
> ..\tp/Texinfo/Report.pm line 244
>      Makefile:1491: recipe for target `texinfo.info' failed
>      make[3]: *** [texinfo.info] Error 7
> 
> How to handle this?  MS-Windows doesn't support LC_MESSAGES in its
> setlocale API, I guess that's what triggers the problem.  (And of
> course Windows also doesn't have a UTF-8 locale.)  I replaced
> LC_MESSAGES with LC_ALL, and the compilation finished successfully,
> but is that the right fix?  What is this stuff used for, and how can I
> test specifically that whatever it's used for works in the built
> Texinfo?

This is used specifically for Texinfo documents that are not in English.
The easiest way to check this is working properly would be with a short
input file like:

----------------
\input texinfo  @c -*-texinfo-*-

@documentlanguage de

@node Top
@top top

Top Node

@node One
@chapter Ein

Das erste Kapitel ist da.

@bye
----------------

then run texi2any --html --no-split TEST.texi.  Check TEST.html to see
that it contains translated strings in output such as

<div class="header">
<p>
Vorige: <a href="#Top" accesskey="p" rel="prev">top</a>, Nach oben: <a 
href="#Top" accesskey="u" rel="up">top</a> &nbsp; </p>
</div>

("Vorige" rather than "Previous").

The code switching to a UTF-8 locale is supposed to work around quite
an irritating problem that I never found a solution I was happy with.
I described the issue here:

https://lists.gnu.org/archive/html/bug-texinfo/2020-11/msg00088.html

I don't think there would be an issue with using LC_ALL instead of
LC_MESSAGES as that is what was done before commit fc28011fb6b34bb
(on 2020-11-29).