branch master updated: tp/texi2any.pl: update comment on DOC_ENCODING_FOR_INPUT_FILE_NAME on Windows (based on Eli input and tests results).

[Patrice Dumas]

This is an automated email from the git hooks/post-receive script.

pertusus pushed a commit to branch master
in repository texinfo.

The following commit(s) were added to refs/heads/master by this push:
     new f76c70737b tp/texi2any.pl: update comment on 
DOC_ENCODING_FOR_INPUT_FILE_NAME on Windows (based on Eli input and tests 
results).
f76c70737b is described below

commit f76c70737b52a0c3e8b1b97731abb51e15371558
Author: Patrice Dumas <pertusus@free.fr>
AuthorDate: Wed Oct 26 00:46:23 2022 +0200

    tp/texi2any.pl: update comment on DOC_ENCODING_FOR_INPUT_FILE_NAME on
    Windows (based on Eli input and tests results).
---
 tp/texi2any.pl | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/tp/texi2any.pl b/tp/texi2any.pl
index fa2cc88c72..f1573241fa 100755
--- a/tp/texi2any.pl
+++ b/tp/texi2any.pl
@@ -332,10 +332,14 @@ my $main_program_set_options = {
     'LOCALE_ENCODING' => $locale_encoding
 };
 
-# use locale on Windows to set encoding of input file name as
-# system calls obey locale pages even if the filesystem uses
-# utf16 internally
-# FIXME better explanation?
+# In Windows, a character in file name is encoded according to the current
+# codepage, and converted to/from UTF-16 in the filesystem.  If a file name is
+# not encoded in the current codepage, the file name will appear with erroneous
+# characters when listing file names.  Also the encoding and decoding to
+# UTF-16 may fail, especially when the codepage is 8bit while the file name
+# is encoded in a multibyte encoding.
+# We assume that in Windows the file names are reencoded in the current
+# codepage encoding to avoid those issues.
 if ($^O eq 'MSWin32') {
   $main_program_set_options->{'DOC_ENCODING_FOR_INPUT_FILE_NAME'} = 0;
 }