When I wrote it last time I wasn’t really planning on making this a particularly regular column, especially since the health of Gentoo is rarely good, and I rant enough about it that just adding one more series of articles is not going to be very helpful.
I haven’t really ranted much about the KDE 4.10 update, on my blog at least, although I did complain about it on Google+ — it wasn’t, though, a problem with Gentoo for the most part, but rather a QA failure within the KDE project coupled with a smaller failure in GnuTLS which caused headaches to just a bunch of users in ~arch, which is not a big deal.
On the other hand, Paweł wrote about the bigger failures for the past year or so (I would say the worst failure in upgrades since we got rid of Arfrever!) — the problem did not lie on the packages themselves, if upstream decides to change the rules of the game, it’s their prerogative, but we should have handled them properly, among others, by releasing news items for the both of them.
It’s not like these have been the only screwups, for instance miredo is still broken after ifconfig and ip have been moved to /bin, because Mike decided that adding a compatibility symlink was just a workaround…
the current status of ~arch is actually quite decent. The new Boost is masked (and will likely stay that way for a little while); we got rid of Ruby Enterprise (which was not maintained), and Rubinius really never entered the tree for real due to their choices of not making releases and not following LLVM closely enough, so even that is clean. We still lack a decent, recent JRuby, but the problem there is twofold: I don’t know enough about packaging Java, which is bothersome enough when I’m trying to get epubcheck3 in tree, and there are new dependencies that need to be packaged. Given my job direction, I doubt I’ll have much time for this though.
For what concerns the output of the tinderbox the situation is .. okayish. The stable testing actually entered a lull of looping around the same packages over and over again, which means it can relax for a little while. Until a new stable package is released. The ~arch testing is currently running the reverse dependencies of pkgconf, at least as much as they can hit, to see if it’s possible at all to discuss changing the default. I’m not sure if that’s going to be a good idea to be honest, but I’ll leave that to the rest of the developers to decide.
So anyway, this is the bottom line for the moment, I would say, I’m still hoping for things to improve over time instead of getting worse, but that kind of hope is considering taking a long hiatus — especially since there are build, test, or QA failures that haven’t been worked on in over three years…