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

On Sun, Jan 18, 2026 at 06:31:07PM +0200, Eli Zaretskii wrote:
> > Date: Sun, 18 Jan 2026 16:49:38 +0100
> > From: [email protected]
> > Cc: [email protected], [email protected]
> > 
> > On Sat, Jan 17, 2026 at 12:40:07PM +0200, Eli Zaretskii wrote:
> > > AFAICT, the only differences are:
> > > 
> > >   . I get diffs, whereas you don't
> > >   . in my case the command says TEST=1, not TEST=2
> > >   . the exit status is 1 and not 0
> > > 
> > > Does this give any clues as to what is going on?
> > 
> > Not really.  If TEST=2, there are additional checks of reference count
> > of Perl objects, you probably do not have the modules needed by that.
> > It is not important, the bugs that can be found like that are not
> > important bugs and probably not platform dependent.  So TEST=1 is ok,
> > and should not interfere with this test.
> > 
> > Maybe something that could be done easily to check the file name used
> > for the Info file would be to call texi2any.pl with --verbose, so like:
> > 
> > /d/usr/Perl/bin/perl -w ./..//perl/texi2any.pl  --force --conf-dir ./../perl/t/init/ --conf-dir ./../perl/init --conf-dir ./../perl/ext -I ./formatting -I formatting/ -I ./ -I . -I built_input -I built_input/non_ascii --error-limit=1000 -c TEST=1 --verbose --output formatting/out_parser/reuse_macro_expand_file/ --macro-expand=formatting/out_parser/reuse_macro_expand_file/simplest.info ./formatting/simplest.texi
> > 
> > In my case, I get this information:
> >   Output file formatting/out_parser/reuse_macro_expand_file/simplest.info
> > 
> > If it is not the same, it could explain why there is no 'overwriting
> > file' warning.
> 
> I see this:
> 
>   Output file formatting\out_parser\reuse_macro_expand_file/simplest.info
> 
> So something, probably Perl I'm using (being a Windows port of Perl),
> outputs file names with backslashes, and then some other code probably
> compares file names as simple strings.  Could that be the reason?

Yes, it is the reason.  What is unclear to me is where the backslashes
come from.  I do not imagine Perl changing the line command arguments,
but I may be wrong.

We tried to at least avoid adding backslashes all over texi2any code, so
hopefully it is not added by us.  In addition the C parser is used and
I can't imagine that we add backslash.

Could you please apply the attached patch, and rerun the same command as
above, and report the output, please?  I added some printout of the
directories name to try to narrow down where the backslash appear.

-- 
Pat

diff --git a/tta/perl/Texinfo/Convert/Converter.pm b/tta/perl/Texinfo/Convert/Converter.pm
index 3a7e5e3868..73d33d8ed2 100644
--- a/tta/perl/Texinfo/Convert/Converter.pm
+++ b/tta/perl/Texinfo/Convert/Converter.pm
@@ -1286,6 +1286,7 @@ sub determine_files_and_directory($$) {
     }
     if (defined($self->get_conf('SUBDIR')) and $output_file ne '') {
       my $dir = File::Spec->canonpath($self->get_conf('SUBDIR'));
+      print STDERR "DFD SUBDIR: '$dir'\n";
       $output_file = join('/', ($dir, $output_file));
     }
   } else {
@@ -1339,7 +1340,10 @@ sub determine_files_and_directory($$) {
     $destination_directory = $output_dir;
   }
   if ($destination_directory ne '') {
-    $destination_directory = File::Spec->canonpath($destination_directory);
+    my $canon_destination_directory
+      = File::Spec->canonpath($destination_directory);
+    print STDERR "DFD dir '$destination_directory' '$canon_destination_directory'\n";
+    $destination_directory = $canon_destination_directory;
   }
   return ($output_file, $destination_directory, $output_filename,
           $document_name, $input_basefile);
diff --git a/tta/perl/texi2any.pl b/tta/perl/texi2any.pl
index 90331ff8c0..f68ae21295 100755
--- a/tta/perl/texi2any.pl
+++ b/tta/perl/texi2any.pl
@@ -1789,6 +1789,7 @@ while (@input_files) {
   } else {
     $canon_input_dir = File::Spec->canonpath($input_directory);
   }
+  print STDERR "III '$input_file_name' '$input_directory' '$canon_input_dir'\n";
 
   #my $input_file_base = $input_file_name;
   #$input_file_base =~ s/\.te?x(i|info)?$//;