On 22/03/2015 20:08, lee wrote: > Jonathan Callen <jcal...@gentoo.org> writes: > >> On 2015-03-22 09:04, lee wrote: >>> Tom H <tomh0...@gmail.com> writes: >>> >>>> On Sun, Mar 22, 2015 at 7:31 AM, Alan McKinnon >>>> <alan.mckin...@gmail.com> wrote: >>>>> >>>>> It looks to me like sysvinit-2.88-r7 was stabilized and the >>>>> maintainer of apcupsd didn't notice. From the ebuild for >>>>> apcupsd-3.14.8-r2: >>>>> >>>>> DEPEND=" || ( >=sys-apps/util-linux-2.23[tty-helpers] >>>>> <=sys-apps/sysvinit-2.88-r4 sys-freebsd/freebsd-ubin ) >>>>> >>>>> What I suggest is copy that ebuild to your local overlay and >>>>> update the DEPEND to <=sys-apps/sysvinit-2.88-r7 and redigest >>>>> >>>>> If that gives a correct update path for world, then file a bug >>>>> against apcupsd. >>>> >>>> Some commands were moved from sysvinit to util-linux, and these >>>> commands are required by apcupsd and are included in util-linux >>>> if it's compiled with the "tty-helpers" use flag. >>> >>> Is this somehow reflected in the dependencies? And how could I >>> deal with the multiple versions of util-linux that seem to be >>> required? >>> >>> Perhaps I should forcefully update util-linux and use tty-helpers >>> so that apcupsd still works in case I reboot. But what other >>> problems might that cause? >>> >>> >>> What am I supposed to think? Should we not update unless no >>> problems are listed and just wait in case there are some, >>> potentially having to wait indefinitely? How about security >>> updates then? >>> >>> >> >> It is reflected in the dependencies by the fact that the first dep >> (and generally the one chosen by portage) requires a new-enough >> version of util-linux *with the "tty-helpers" USE flag enabled*. You >> don't need multiple versions of anything installed. If you just add >> "sys-apps/util-linux tty-helpers" to your /etc/portage/package.use >> file and try again, you will likely find that portage will update >> everything for you without any further issues. > > Oh that actually works! How is one supposed to know that this use flag > must be added?
Sadly, you don't know. There is no clue in any of the output you posted that this is required, so your only solution is to ask the collective memory of the community. Lucky for you and others, Jonathan was aware of the problem and was kind enough to post the solution. This is one of the things that is starting to real get on my damn tits about portage, for about 2 years now. It's not an easy problem to solve, and to be honest, portage is not helping at all. You have two options in running it: don't use -v and get very little info, or use -v and get a terminal dump of the entire graph tree with lots of stuff and zero real information about how to solve it. Look at my thread with Dale just the other day, I managed to help him with the correct answer because I had a magic brainwave to search for the "<" character. Seriously, what kind of process would ever use that as a problem solving approach? In your case, the solution is in the ebuild for acpupsd and it's specific DEPENDs. Now, I'm generally OK with looking in ebuilds for real answers and have gotten used to it, but ffs I should not have to do that. Well-written software should provide that information in it's output, and it shouldn't be hard to get the software to do it. Ok, rant over. -- Alan McKinnon alan.mckin...@gmail.com
