Alternatively, we could run dos2unix over all the tests during configure stage and error if this makes a diff. https://linux.die.net/man/1/dos2unix Best regards, Jacob Faibussowitsch (Jacob Fai - booss - oh - vitch)

On Oct 28, 2021, at 08:22, Jacob Faibussowitsch <[email protected]> wrote:

Fix is trivial, we just need to replace ‘\n’ with os.linesep everywhere.

https://docs.python.org/3/library/os.html#os.linesep <https://docs.python.org/3/library/os.html#os.linesep>

Best regards,

Jacob Faibussowitsch (Jacob Fai - booss - oh - vitch)

...

On Oct 28, 2021, at 08:12, Matthew Knepley <[email protected] <mailto:[email protected]>> wrote:

I just found a loophole in the test parsing happening here:

https://gitlab.com/petsc/petsc/-/blob/main/config/testparse.py#L530 <https://gitlab.com/petsc/petsc/-/blob/main/config/testparse.py#L530>

If your file was produced on Windows (CRLF), then the parsing fails and tests are silently ignored. This took me some time to find and could be confusing for others. I think we should either:

a) Fix the parser to work with CRLF as well

or

b) Fail immediately for CRLF files

What do you think?

Thanks,

Matt

-- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>