On Sat, Mar 27, 2010 at 12:57:24 +0100, Reinier Lamers wrote: > > - We cannot guarantee correctness in case of untagged patches. > > - But we can guarantee it on tagged patches. > > No, I don't, actually. You can agree on the correctness for a meaningful > subset of untagged patches - namely those created by a UTF-8-aware darcs. > That's the most important. For the rest we'll have to guess anyway.
Oh! Between the two of you, I think I'm starting to get it... true positive [UTF-8 detected and it really is UTF-8] This happens for patches created by new Darcs or by old Darcs in a UTF-8 based locale false positive [UTF-8 detected but it's not really UTF-8] Possible but rare! -- Will never happen with new patches because they are all UTF-8 encoded true negative [UTF-8 detection fails and it's not a UTF-8 patch] This would happen with some old patches false negative [UTF-8 detection fails but it really is a UTF-8 patch] Also impossible by definition > The standard behavior, afaik, with data of which you don't know the encoding, > is to treat it as encoded in the current locale. Tagging newer patches is the > price we pay for that. Apparently, many people don't want to pay that price. > And even without tagging, things will only go wrong in rare corner cases > where > something is valid UTF-8 and valid in the locale at the same time, and the > locale is not UTF-8, and the locale and UTF_8 actually differ in their > interpretation of the metadata. In the rare corner case, as Petr says, we can only make a guess, and [I add?] that guess would only be right if the person looking at the patch is in the same non-UTF-8 locale as the creator. Phew. Have I got it now? As long as you and (your reviewer) are happy, I'm happy :-) -- Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow> PGP Key ID: 08AC04F9
pgp5N0uEvrbhF.pgp
Description: PGP signature
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
