Re: Texinfo 7.2.90 build with mingw.org's MinGW: testing [info]

> Date: Fri, 16 Jan 2026 18:03:04 +0200
> From: Eli Zaretskii <[email protected]>
> Cc: [email protected], [email protected]
> 
> > From: Gavin Smith <[email protected]>
> > Date: Thu, 15 Jan 2026 18:58:38 +0000
> > Cc: [email protected], [email protected]
> > 
> > On Thu, Jan 15, 2026 at 08:25:14PM +0200, Eli Zaretskii wrote:
> > > I guess I'm missing something, then, because the test script does
> > > this:
> > > 
> > >   srcdir=${srcdir:-.}
> > >   . $srcdir/t/Init-test.inc
> > > 
> > >   # Follow an reference to "file-menu" in dir without matching label exactly
> > >   $ginfo --output - FiLe-M | grep "^File: file-menu,"
> > > 
> > > I understand that the Info file it uses is infodir/dir, is that
> > > correct (even this was something it took a lot of time to understand)?
> > > But then what is "FiLe-M" in the above command?
> > > 
> > > Thanks.
> > 
> > Yes, it loads infodir/dir initially.
> > 
> > The "FiLe-M" should use the following entry in that file:
> > 
> > * file-menu: (file-menu). 
> > 
> > The test is to check if it can still find that dir entry even though the
> > case of the letters does not match.
> > 
> > It could be related to multibyte support as the comparison is done
> > with the mbscasecmp function from gnulib.
> 
> No, it's much simpler and sillier: the problem was with
> manual-not-found hook.  That file is a Unix shell script, which
> MS-Windows 'popen' cannot run, because the Windows shell cmd.exe
> doesn't know how to run Unix shell scripts.
> 
> So we need to have a Windows variant of manual-not-found hook, and
> we need to look for it when running on Windows.
> 
> In addition, the !PIPE_USE_FORK branch of get_output_from_program had
> a few problems: it failed to use the FILENAME argument when
> constructing the command line for popen, used the null
> formatter_args[2] in sprintf, and had a few other minor issues.
> 
> Here's the patch to fix all that, after which all Info tests pass.

Gavin, is there some problem with the patch I sent to fix this?  I see
that it was not yet installed in Git.  Is there anything I can do to
help?