On Sun, Oct 04, 2026 at 01:56:05PM +0200, Patrice Dumas wrote:
> As far as I can tell, we don't.  What we do is that we skip some tests
> in some cases, but it is not ideal, because the regeneration of the test
> results is not done and there can be inconsistencies.

Answering to myself, having comparison independent of the end of line
allows to have the tests proceed, but regenerating test results on a
native Windows platform would lead to replacing files with LF with files
with CRLF, something we do not want.  We do not want it in version
control, but we do not cant that in distributions either.  This is an
issue both for tests in tta/per/t and tests in tta/tests.

There is no urgency, as somebody developping texi2any in native Windows
is probably very uncommon, but it would be good to have it fixed anyway.

I am not actually sure about the best way to fix that, but I'll try
something.  A difficulty is that I cannot easily test that kind of
change.

-- 
Pat

Reply via email to