I commented out a use flag, but portage didn't bother. Had to touch the ebuild, to get the workdir populated with fresh sources.
The USE= part of `emerge info` is pretty useless too, because it may not print all the use flags used by a particular ebuild.
WTF are you talking about ? Please be a bit more specific as - portage.use (from the summary) doesn't mean anything to me - `emerge --info` has nothing to do with any particular ebuild, it's system state information - when/where/how did you comment the use flag in what ebuild ?
Sorry, I meant /etc/portage/package.use of course! comment #1 is either another issue or a non-issue, that's why I didn't opened another bug report. I know that it's system state information, but when a user reports a problem with a particular ebuild, shouldn't the use flags be known, which are actually used? E.g. something like "emerge info <ebuild>"
emerge -pv <ebuild> Is it what you want?
Masamoto: More or less, a dev usually just asks for emerge info and not emerge -pv, so a dependency related problem may be overlooked in the first place. But please forget comment #1. I'll outline the reason why I opened the bug: A precondition is having noclean in features, I think. 1. I merged the latest mozilla ebuild, just to have a look if the acrobat/gtk2 problem is solved. It isn't. 2. I put "net-www/mozilla -gtk2" in package.use and merged it again Portage detects now, that the workdir is fine and everything is compiled, so it installs the +gtk2 version, without taking notice that I changed the use flag in package.use in between
You asked it not to clean... So it didn't... Thus it "breaks". The reason I see for it being "wrong" is that on a full, normal emerge, it sould always clean first, including the environment file. You didn't specify whether you ran emerge or ebuild.
I ran emerge, but I usually have "noclean" in features end clean the directory every now and then by hand. I would really prefer it, when use flag changes would be checked, before reusing the existing workdir/environment. It would make it easier to test different use flag combinations. I always assumed portage would do this with USE= in make.conf, but your reply sounds like it doesn't. If you think this would add too much complexity, then drop this request, please.
Just noticed, that I'm not the first one, who requests this. *** This bug has been marked as a duplicate of 2427 ***