Hi guys,

I've also been asked to take a look at this issue, but my time is still
a bit iffy and I'd be more comfortable with Ganesh doing it if he's
available.

That said, I'll contribute some minor stuff just tackling the fringes
of this patch.

I noticed that there was some uncertainty about what checkpoint patches
were.  Even though this is likely redundant and outdated, I've begun
noting some stuff down on http://wiki.darcs.net/Checkpoints.  Hopefully
others who know about checkpoints or learning about them can jump in.

I don't think checkpoints are anything particularly scary.  As far as I
understand they are just the mechanism by which darcs get --partial
works.  I think a checkpoint patch is just a patch that has the same
cumulative effect as a set of patches under a given tag.  That's all
there is to it.  Rather than fetching and applying a bunch of patches,
you apply the one checkpoint patch that leaps from point A to point Z
without any of the intervening changes.  For example, if you had named
patch 1: move f g, followed by named patch 2: move g h, the checkpoint
patch created for it would be 'move f h'.  It's a giant coalescing
operation!  If you have a copy of old darcs lying around, try making a
checkpoint patch and see.  In fact, make two in succession so that you
can see how one checkpoint builds off the last state.

So we're no longer creating new checkpoints from Darcs 2.3 on.  The
question I wanted to ask myself in this leg of the review is if we do
away with the code in Darcs.Repository.DarcsRepo that manipulates the
checkpoint inventory.  I guess the only reason this code would be useful
is if you were to either unpull/obliterate a tag.

Perhaps that provides some decision-making context?

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9

Attachment: pgpXd5hZ0p7tc.pgp
Description: PGP signature

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

Reply via email to