Re: patch --truely-rejected

[Akim Demaille]

>>>>> "Paul" == Paul Eggert <address@hidden> writes:

>> BTW, I applied the Peter Eisentraut's patch after yours was
>> installed, and many bits were common.  Unfortunately, I could not
>> find a means to have patch create a `.rej' with *only* the bits
>> which are (apparently) not installed.  Would that make a easy
>> addition to patch?

Paul> Sorry, I don't quite follow, since I think 'patch' already does
Paul> what you want.  I.e., 'patch' already puts into the .rej file
Paul> only the hunks that are not applied.

Sorry, I'm not clear at all indeed.  I was asking whether it would be
possible for patch to keep only the chunks that 1. could not be
applied, 2. are not apparently applied.  Currently it puts in .rej
those matching 1 independently of 2.

When you apply twice a patch, patch complains saying that the patch
was apparently already applied, so it seems that it knows how to check
whether the expected result is already in the file to patch.  If it
could keep/reject bits to leave in .rej based on this, you'd get the
chunks which are ``really not applied''.