On Sat, Nov 29, 2008 at 12:23:47PM +0000, Ganesh Sittampalam wrote: > On Sat, 29 Nov 2008, Trent W. Buck wrote: > >> | Tagging also provides significant performance benefits: if Darcs sees >> | that two repositories both have a tag, it can safely assume that both >> | repositories also have all the patches `in' that tag. When a tag >> | covers thousands or tens of thousands of patches, this saves a lot of >> | time. > > The situation is actually slightly more complicated; this is only > true if the tag is "clean" in both repos - where clean = "the only > patches before this tag are the ones that it depends on (directly or > indirectly)". But since I doubt the documentation wants to go into > that level of detail, or deal with the issue of ordering of patches > in a repo in too much detail, you could just say something like "it > can often safely assume" or "it can usually safely > assume". Unsatisfyingly vague, but more accurate.
Well, IIUC *technically* the assumption I gave is still accurate, even if the code doesn't actually do it quite that way :-) I'll working on rephrasing it to match the description you gave above. _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
