On Sat, Sep 04, 2010 at 15:17:26 +0000, Petr Ročkai wrote: > Thu Aug 12 01:30:27 CEST 2010 Petr Rockai <m...@mornfall.net> > * Merge Darcs.RepoPath into Darcs.Path.
Petr later worked out why we have SubPath and not Relative (which I'd completely forgotten about in my review, oops). The point of SubPath is that it has no initial ".." (or anything that could lead to one) > * Drop unused and redundant pathFromFileName. > * Make the tests pass (mostly ./foo versus foo in user-visible output). Quick look at these > * Optimize darcs diff. Reviewed by time machine > * Resolve hashed-storage dependency conflicts. > * Resolve conflict in Commands.Diff imports. > * Resolve conflicts in Commands.WhatsNew imports. > * Fix conflicts in Darcs.Diff. > * Resolve conflicts in Darcs.Patch.*. > * Resolve a (bogus?) conflict in darcs.cabal. Ignored on the grounds of boringness. :-) > * First round of OF removal. > * Also remove support for getting old-fashioned (to hashed). > * Only test Darcs2 and Hashed (and not OF). Blazed through these as they looked easy > * Set a $format variable in lib, depending on current repo format under > test. > * Make the tests a bit less verbose (do not print all of lib every time). Cherry picked these > * Modernize the conflict-doppleganger test. > * Fix up the module list in darcs.cabal. > * Update couple more tests for absence of OF. Kind of ignored these.... More to go! If anybody wants to help chip away at this bundle, could be good, or maybe we should focus on getting fast-patch-production logistics settled. Drop unused and redundant pathFromFileName. ------------------------------------------- OK Make the tests pass (mostly ./foo versus foo in user-visible output). --------------------------------------------------------------------- > +cat _darcs/patches/pending Quite a few of these debugging statements thrown in, same old yada-yada about separate patches. > -echo 'M ./foo +1' > correct_summary > +echo 'M foo +1' > correct_summary OK so I see that this is main change of the patch. Is this user interface change going to pose a backwards compatibility problem for people who are scrape-parsing our human readable output, and should if so, to what extent should we care? Enough to post an advisory in release notes? Enough to try to preserve existing behaviour? Maybe now that we remember the SubPath/RelativePath distinction, it could be worth revisiting this question. Optimize darcs diff. -------------------- Previously reviewed. I remember pointing out some file move problems in the command line arguments. -------------------------------------------- Resolve hashed-storage dependency conflicts. Resolve conflict in Commands.Diff imports. Resolve conflicts in Commands.WhatsNew imports. Fix conflicts in Darcs.Diff. Resolve conflicts in Darcs.Patch.*. Resolve a (bogus?) conflict in darcs.cabal. -------------------------------------------- Given that I'm reviewing a Review Team patch, I can just assume these are boring and ignore ---------------------------------------------------------------------- First round of OF removal. Also remove support for getting old-fashioned (to hashed). Only test Darcs2 and Hashed (and not OF). ---------------------------------------------------------------------- I blew through these three patches because they look like fairly straightforward machete operation, basically removing if foo then hashedStuff else oldStuff with hashedStuff I'm also going to ignore the problem of how to make optimize --upgrade work again when all the infrastructure has been blown away. Chopping through the jungle, whee! I'm more much enthusiastic about this work because I feel like we've made our attempt to be responsible about it and do our darcs-users homework first: http://lists.osuosl.org/pipermail/darcs-users/2010-August/024941.html > $$ text "the maintainer to > run darcs optimize --upgrade with darcs 2.4.0 or higher?" > $$ text > "*******************************************************************************" Looks like warning text that could go too, but I'll bet you got rid of this later. > -copyRepoOldFashioned :: RepoPatch p => Repository p C(r u t) -> [DarcsFlag] > -> String -> IO () > -copyRepoOldFashioned repository opts repodir = do This one feels particularly nice to be rid of. Checkpoints, partial repositories, a whole lot of mess we can get rid of. > rmfile ./src/Darcs/Repository/Checkpoint.hs > rmfile ./src/Darcs/Repository/Pristine.hs Rah! ----------------------------------------------------------------------- Set a $format variable in lib, depending on current repo format under test. Make the tests a bit less verbose (do not print all of lib every time). --------------------------------------------------------------------------- Cherry picking these patches to the test suite which look like good ideas. Modernize the conflict-doppleganger test. --------------------------------------------------------------------------- Didn't line-by-line this, as it looks like just some refactoring (we do still have advice to avoid functions in test scripts, perhaps worth revisiting) and removing any old-fashioned bits from the test. Fix up the module list in darcs.cabal. -------------------------------------- > - Darcs.Repository.DarcsRepo > + Darcs.Repository.Old Was just a rename in your first round of OF removal patch Update couple more tests for absence of OF. ------------------------------------------- Treating as boring. -- Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow> For a faster response, try +44 (0)1273 64 2905 or xmpp:ko...@jabber.fr (Jabber or Google Talk only)
signature.asc
Description: Digital signature
_______________________________________________ darcs-users mailing list darcs-users@darcs.net http://lists.osuosl.org/mailman/listinfo/darcs-users
