[darcs-users] Re: why is darcs getting the "root" version of a file

[David Roundy]

On Fri, Jul 29, 2005 at 01:28:05PM +0100, Jamie Webb wrote:
> On Fri, Jul 29, 2005 at 06:34:03AM -0400, David Roundy wrote:
> > With a 'pull' we would have announced the conflict.  The trouble is that
> > with a get, checking for conflicts could be very expensive (depending how
> > far back in the history we went--in theory we'd have to go all the way back
> > to the second patch, which would cost us O(N^2) or worse in the total
> > number of patches (depending how unlucky we are).
> 
> Seems like that could be avoided with a simple flag in the inventory
> marking patches as containing unresolved conflicts. Darcs must've been
> aware of the conflict when it first appeared.

Darcs currently doesn't discover (although it could... it'd just be more
code) which patch has a conflict, it just knows that there are conflicts
(and in which files).

We'd also have to add a check to see when conflicts are resolved when we
pull, apply or record.  And when patches are unrecorded, unpulled or
amend-recorded, we'd have to check whether any conflicts had been
unresolved.

I definitely plan on reworking the conflict marking code when the new
conflictors code is ready.  For now, conflict handling is sort of on hold
(or you could say in progress).
-- 
David Roundy
http://www.darcs.net