3 refactorings done while looking into http://bugs.darcs.net/issue2436 .
The third one (make patch selection lazier) actually does a difference
when moving around lots of patches (~19 to ~11seconds when running
"darcs rollback -p ." on darcs' repo),
althought it does not solve the underlying issue of the aforementioned
bug.
3 patches for repository http://darcs.net:
patch 9cfe94d440f65c2891316abbcd50b57706de695e
Author: Guillaume Hoffmann <guillaumh@gmail.com>
Date: Fri Sep 18 08:50:33 ART 2015
* inline patchSetToPatches in the only place where it is used
patch 03b925ae81a3499c447dbe53954cb472c8894483
Author: Guillaume Hoffmann <guillaumh@gmail.com>
Date: Sat Sep 19 18:33:59 ART 2015
* get rid of selectChanges
patch 4d55c7ccc1e8cbd80066c05d66e6309bd96aed8e
Author: Guillaume Hoffmann <guillaumh@gmail.com>
Date: Sat Sep 19 18:36:48 ART 2015
* make patch selection lazier in presence of matchers
I would self accept the bundle but I kind of regret the patch "get rid
of selectChanges" now that I remember that Florent did some work on
SelectChanges two years ago which was never merged (see
http://blog.darcs.net/2013/02/darcs-hacking-sprint-8-report.html ). I
wrote to him about it and am waiting for his answer.