Cleaning up the digiKam docs mess

With the recent discussion about Google’s Picasa port to Linux, digiKam got its share of attention, being, for many people, a good alternative to Picasa, being Libre and native. And if you would consider installing Wine, even if you’re a GNOME user you should be ready to install KDE, too 🙂

But with this attention, there was again whining about the ~40MiB digiKam documentation tarball. The problem is that it is all user documentation, but with 19MiB of images to make it simpler to understand, and in the past the solution was “download everything, it’s user doc, you need it”, but this of course does not scale at all for dialup users or users that pay per traffic.

Today I reached an agreement with Mark (as representative of QA Team) to make use of the doc useflag for media-gfx/digikam to download this extra pack of documentation, as it requires time and a lot of downloads. It might be “out of standard” with respect of other KDE ebuilds, but I care about standards only when they make sense to me, and forcing 10 * $package_size MiB of extra downloads does not make sense to me.

To try limiting the impact even when using documentation, I’ve re-packed digiKam’s doc, so that all the languages are in a different tarball, to just have the English documentation you’ll download 19MB, and that plus something more if you’re enabling other languages, you’ll never get more than 30MiB for any language, it’s one hour less on 56k!

The problem now is testing the ebuild as it’s a long change.. and as carlo does not seem to like this approach, I’ll be probably the only one maintaining digiKam from now on, which means that it might take some more time to get it bumped and stuff (also because I have to split the tarballs).

I’m not sure if I’ll have time tomorrow, but absolutely in the next days I’ll write a maintainer’s guide for digiKam describing the problem, the approached solution, and the script, telling what it’s needed to run it. I have to thanks again the guys of the AMD64 team, as I’m letting pitr do the hard job for me (also because that makes it take just seconds to put the data on the mirroring directory 😉 ).

I’m spending quite a bit of time on this, unfortunately :/ but hey, I like when my users are happy 😉 at least they won’t hate me 🙂