Op wo 07-01-2004, om 16:34 schreef Dominique Devriese: > Colin Watson writes: > >> OK, the problem of filing bugs on every individual package, and > >> praying they get accepted, remains. > > Don't pray that they get accepted. Take the initiative and NMU in > > the usual way after some suitable period. Everyone else doing > > general work on our packages has to do exactly the same thing; what > > gives translators a different status? > Basically because there is a better alternative for translators. Such as? I've seen a proposal, but it's not a "better alternative". > For > other people doing "general work", there often isn't any. If you > think there is, feel free to propose that as a separate proposal, > instead of trying to discourage this one. > > I mean, how can you really see this as a valid argument against the > proposal. I have indicated real and fundamental problems with the > current approach, a new system, with real and proven advantages, Sorry, but that's not true. KDE != Debian. For one thing, Debian works with package uploads; it does not use a central CVS server as the KDE project does. That's more than just an implementational detail, it's a real organizational difference, which has impacts on many areas of how we work. For another, it's not unreasonable for the KDE project to assume that a user would install the entirety of KDE on his system, or at least that he would install most of it; so having a few unused translated strings on a system is not really a problem. The same cannot be said for the Debian distribution -- it's not even technically possible to install everything should one want to. Therefore, providing a package with translated debconf messages will put our users before a dilemma: either they install a huge package with translated strings of which they will perhaps use 10% (having the remaining part of the package use up valuable diskspace), or they can forget about translated installations. This isn't true when you're considering translations to different languages being shipped in the same package; there's localepurge to fix that issue. Your suggestion does have proven advantages, but they apply to a completely different environment which can in no way be compared to Debian. > and > all you say here is "I don't like to change. Furthermore, in > completely different situations, we also use NMUs". Please provide > arguments based on advantages and disadvantages of the proposed > system. Those arguments have been given. Two examples: there are dependency problems, and the approach discourages a maintainer to perform QA on the provided translations. What Colin says here is that in Debian, the canonical way of doing work on a package of which you are not the maintainer is to do an NMU of that package. Creating per-language packages is an option, but it has a price, and the advantages do not necessarily outweigh its disadvantages; therefore, I think Colin suggests (and I agree) that translators should use NMU's when their contributions aren't incorporated. After all, that's what NMU's are for. -- Wouter Verhelst Debian GNU/Linux -- http://www.debian.org Nederlandstalige Linux-documentatie -- http://nl.linux.org Most people have two reasons for doing anything -- a good reason, and the real reason
Attachment:
signature.asc
Description: Dit berichtdeel is digitaal ondertekend