> There's nothing like a good minimal test case. :-) > [I assume it's the last darcs record where we hit 'e', right?]
Right.
> But that's not particularly surprising if we're working on the principle
> that the hunk editor does not change the working directory. In effect
> you've edited the patch into oblivion, so Darcs does not propose
> anything; but it does propose the "anti-patch" afterwards which brings
> back the hunk. This is effectively a no-op.
Ok, I can see how that's technically consistent then, but it's just not
what I expected.
My common use-case for this is that I have code I want to record, and
debugging I want to remove. The current implementation is still useful
for this, because it would allow me to split a case where the debugging
touched the code.
If I can do that, I suppose I can just do a "revert -a" when I'm done to
eliminate the debugging code that remains.
> Also, unless I'm missing something, wouldn't you have just said 'n'
> in the traditional interface for this case?
Right, and then later a "revert". I was hoping to get two steps down to
one. Instead of record+revert, I could just do a "record".
It will still be an improvement over the the worst case:
record
# oops! manually delete debugging to split a hunk
record again
revert the rest
Mark
signature.asc
Description: PGP signature
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
