On Wed, Mar 30, 2005 at 01:13:12AM +0200, foser wrote: > On Tue, 2005-03-29 at 16:29 -0600, Brian Harring wrote: > > If you use the approach I've laid out (yes, not new, I laid it out in > > 24439) you wouldn't have to dick around with deprecating a version, > > nor essentially mandating what version is default. You'd leave the > > total control over what versions the user wants to deal with in the > > hands of the _users_, and what versions the package supports would > > be represented properly/clearly in the IUSE. > > I know the proposed system obviously and it's still flawed in that > really doesn't deal with USE flag versioning in a consistent, > predictable way. State how it's flawed please. If you're going to argue this is a repeat of the gtk2/gtk fiasco that's in the tree currently, please validate how/why it is. The approach detailed above, and in aug '04 *is* the most flexible approach with use flags that addresses user needs, and is extensible (gtk3).
> > So far... I've not really heard a good reason aside from "it's in > > place, we'll just deprecate gtk v1 instead of clean it up" for why > > this cannot be corrected _now_, or really in the past. > > That is all the reason needed here. I'm no biggie on the current > situation (altough if people actually read the USE flag descriptions it > wouldn't be half the issue), but interchanging to be deprecated > behaviour with something just as bad is a regression. I'll post more on > some changes that I'd like to see implemented for this later. > > And the issue never has been about changing a few occurances in the > tree. Well, the issue is what then? It's an often requested change. While people *should* know wth they're flipping on via looking in use.desc, use flags *should* be relatively sane/clear in their implications, at least on a general scale. ~brian
pgpqmr2WD9cbF.pgp
Description: PGP signature
