D2573: xdiff: remove patience and histogram diff algorithms

quark (Jun Wu) phabricator at mercurial-scm.org
Sat Mar 3 21:21:12 UTC 2018


quark added a comment.


  Googling "patience diff", most results will say it is slower but has better diff quality sometimes.
  
  That is very misleading - the algorithm adds greediness (i.e. incorrectness) and is in theory faster in some cases. The "better" quality is also untrue comparing to indent heuristic.
  
  Technically, I don't think it's worthwhile to keep - definitely worse quality and unpredictable performance changes comparing to default diff + indent heuristic setup.
  
  From a non-technical point of view, I think educating people that they cannot be misled is very important. This one alone is an enough reason to do the cleanup, imo.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D2573

To: ryanmce, #hg-reviewers, indygreg
Cc: quark, indygreg, mercurial-devel


More information about the Mercurial-devel mailing list