On Sun, Oct 23, 2022 at 11:45:15AM +0100, Gavin Smith wrote: > With your setup, it breaks at the very beginning, in the Makefile > rule. It would seem to be simple to eliminate the reference to > "included_latîn1.texi" from the Makefile. Then we have to make > sure the tests are skipped.
I eliminated the reference to "included_latîn1.texi" from the Makefile in https://git.savannah.gnu.org/cgit/texinfo.git/commit/?id=d9a947dd76999d0f84499f4cac048363957305bc However, I do not know if the corresponding tests will be skipped, My guess is that the recoding to latin1 will succeed, but the copy of the file may fail, I don't know. But maybe we should try to see if the tests fail or not and do it another way if the tests are not skipped, maybe not having thoses tests run with make check but with specific checks only. Also there is another accented file name in UTF-8 in the Makefile.am in EXTRA_DIST and in the tarball, included_akçentêd.texi: EXTRA_DIST = run_parser_all.sh parser_tests.sh \ $(one_test_files_generated_list) coverage_macro.texi included_akçentêd.texi \ input/included_latin1.texi I guess that 'make dist' on Windows or in any situation in which make does not consider the encoding of the Makefile to be UTF-8 could fail because of that, even before running the command that uses EXTRA_DIST are run. It could also fail when the commands are run too. -- Pat
