Quoting "Stephen J. Turnbull" <[EMAIL PROTECTED]>:

   Clive> and might cause issues for people who have already pulled
   Clive> those two patches.

Indeed it will.  Besides the certain conflicts they'll run into the
next time they pull from your repository, if for some reason they
should publish their repositories, they will be subject to the same
unreasonable demands, with even less justice than in your case.

I looked at the PatchInfo code to generate the hashes which become the filenames for the patches, and I was a bit baffled at what exactly it intends to accomplish. My understanding (which I've verified) is that the author, the patch name, the patch log message, the date, and whether the patch is inverted are all hashed to create the unique filename. Therefore, you can't even edit the log message or the name after recording (and look how many typos there are in darcs' own "changes"!)

It would seem to *me* to be more sane to hash the patch file's contents plus whether it was inverted instead. This way unpull/pull or unrecord/record will only corrupt the repository (from the perspective of other pullers) in the case where a previously pulled patch's _contents_ were actually changed.

(It occurs to me now that in some cases, the same patch name could occur twice in one sequence. touch foo; darcs add foo; darcs record -m 'f'; darcs remove foo; darcs record -m 'f'; darcs add foo; darcs record -m 'f'. I imagine this could be resolved.)

Just some food for thought.

--
Jason M. Felice
Cronosys, LLC <http://www.cronosys.com>
216-221-4600 x302

_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users

Reply via email to