* Eric Blake ([email protected]) [20101215 18:15]:

> This issue has come up in the past.

I though so :)

> To make it work, we'd need an escape sequence \r (we already have escape
> sequences for \n and \\) to uniquely identify whether a trailing carriage
> return is part of the filename or part of the line ending,

Which would be rather easy to do,

> as well as a way to tell whether md5sum is parsing a file generated by the
> new version that knows how to generate \r escapes or the old version that
> blindly output literal \r.

Which wouldn't be so easy AFAIS.

> (using d2u on the checksum file before handing it to md5sum is usually a
> decent workaround to having to teach md5sum about alternate line endings).

That's what the person who asked is doing atm.

Philipp

Reply via email to