[darcs-users] Replacing the n^2 algorithm.

[Anthony Towns]

Ketil Malde wrote:
> Anthony Towns <aj at azure.humbug.org.au> writes:
>>N is the number of conflicting patches.
> It seems to me that a simple (albeit inelegant) way to address *my*
> problem (when the patch can be easily resolved manually) would be to
> have refusable patches.  So when pulling, when I'm asked about the
> Id-removal patch, I can refuse it, and darcs would make note of this
> and never bother me with it again.  

Arguably that'd be a good idea anyway, so that you could pull from a 
development repo to a bugfix repo and be able to say "no" to feature 
additions just once. A user interface like:

$ darcs pull

Tue Jan 11 22:24:39 EST 2005  David Roundy <droundy at abridgegame.org>
   * fix bug in commuting secondary conflictors.

CONFLICTS WITH:
Fri Jan 14 20:04:37 EST 2005  Anthony Towns <aj at azure.humbug.org.au>
   * alternate conflictors fix

Shall I pull this patch? (1/1) [yneWvxqadjk], or ? for help: e

...where "e" (or some other letter) stands for "never" could work, maybe.

> Perhaps it could also note which patch(es) it conflicted with, and not
> let any child repository pull both?  Or let me mark manually two
> patches as mutually exclusive?

I don't think it could extend to child repos; but would it need to? If 
I'm tracking "stable", and suddenly try pulling from the development 
branch, why shouldn't I get prompted for all the new features that've 
been skipped?

(Assuming you can eventually resolve the conflict in the development 
branch, of course)

Cheers,
aj