On Thu, Aug 14, 2014 at 2:26 PM, Alan McKinnon <alan.mckin...@gmail.com> wrote: > On 14/08/2014 18:09, Mike Gilbert wrote: >> On Thu, Aug 14, 2014 at 11:57 AM, Сергей <protsero...@gmail.com> wrote: >>> I have looked at dev-libs/libgamin-0.1.10-r4 and >>> dev-libs/libgamin-0.1.10-r5 ebuilds and compared them. >>> dev-libs/libgamin-0.1.10-r5 has PYTHON_TARGETS="python2_7" (r4 had no >>> PYTHON_TARGETS) and now python-updater doesn't rebuild libgamin. Seems >>> like now everything is ok and it was only a portage bug. >>> >> >> It is actually a bug with python-updater. However, we have no plans to >> fix it; instead, the problem will be resolved once all python-based >> ebuilds are migrated to python-r1.eclass and therefore utilize >> PYTHON_TARGETS. >> >> At that point, python-updater will become obsolete and you will no >> longer need to run it. > > > Um, yeah. > > That's what they said about revdep-rebuild when @preserved-rebuild hit. > And then again when sub-slots hit. But revdep-rebuild to this day still > catches things both of those solutions missed. > > In Gentoo-land I have learned to be extremely wary of any statement like > "old xyz tool is no longer necessary" :-) > > It seems like the 98%-2% rule is still very much in play >
I have not run revdep-rebuild in over a year. If you have seen that preserve-libs is missing things, that's a a bug. Slot-operators are going to take a LONG time to get implemented tree-wide, and I agree that it may never happen. Packages which utilize PYTHON_TARGETS do not get rebuilt by python-updater anyway -- it explicitly skips them. At this point, the majority of packages that people actually use have been converted. Many users may not even need to run python-updater if they don't have USE=python enabled globally.