quark added inline comments. INLINE COMMENTS
> yuja wrote in revset.py:129 > `subset & xs` should be correct since `dagrange` doesn't have > its own order unlike `rangeset`. > > Most revset functions "follow" the default order even if they > are used where they may "define" order. For `subset & xs` to be correct, `subset` needs to be in ascending order. That is true currently. But it is not very obvious why `subset` is in ascending order here (or, the question is, who is responsible to sort it?). I think it's simpler to not depend on it and make every revset respect `defineorder` explicitly. That also allows us to remove some unnecessary sorting. > yuja wrote in revset.py:893 > IIUC, `followorder` is correct because the ordering flags of > `x and y` are flipped as if they were `y and x`. In this case, `y` is expected to completely redefine the order. So `y`'s `subset`'s order does not matter. > yuja wrote in revset.py:1830 > Can you split this to new patch, and preferably include a micro benchmark? > > Revset had historically lots of subtle ordering bugs, and I believe > there are still some. Fewer "if"s should be better in general. I can do that. REPOSITORY rHG Mercurial REVISION DETAIL https://phab.mercurial-scm.org/D451 To: quark, #hg-reviewers Cc: yuja, mercurial-devel _______________________________________________ Mercurial-devel mailing list Mercurial-devel@mercurial-scm.org https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel