On Fri, 20 Jan 2012 11:00:52 -0600 »Q« <[email protected]> wrote: > On Fri, 20 Jan 2012 12:06:22 +0000 > Neil Bothwick <[email protected]> wrote: > > > I suspect this is > > specific to the KDE ebuilds (or eclass). changed-use should only > > skip an ebuild with changed flags if re-emerging would produce > > exactly the same code as before, this may not be the case. For > > example, in some ebuilds, it is the absence of a USE flag that > > triggers an extra configure option, so removing that use flag would > > give the same code as if the package had been emerged with it > > enabled. > > > > Something like this happened recently with the nls flag on glibc. > > Ah, that makes sense -- thanks. (And now I wish I'd read the entire > thread in dev before I posted a few minutes ago.) > > IMO, the man page's section on --changed-use should say what you've > just said rather than what it says now.
I saw the logs from the kde team meeting where they discussed removing kdeenablefinal and decided when to do it. My own update on that ran overnight so I wasn't too fazed. But is this not a case where the kde eclass *explictly* set the USE flag off? (Disclaimer: haven't read the eclass). In that case portage would not know what to do when the flag goes away so the behaviour we saw would not really be a bug, it would be "playing safe" --changed-use is intended for cases like a flag you are not using at all goes away. Caveat: Even then it could still break in subtle ways with dodgy ebuilds. Caveat emptor. -- Alan McKinnnon [email protected]
