As of 2012-03-15, the testing ebuild labelled 20120119_p3256 actually builds the version of 30 November 2011. The 99999999 ebuild builds exactly the same version.
(In reply to comment #1) > As of 2012-03-15, the testing ebuild labelled 20120119_p3256 actually builds > the version of 30 November 2011. The 99999999 ebuild builds exactly the same > version. please provide emerge --info, build logs, etc. thanks
Created attachment 306739 [details] Log of latest unstable build Built with /etc/portage/packages.keywords including the line: sci-electronics/kicad ~x86
Created attachment 306743 [details] Log of bleeding edge build Built with /etc/portage/packages.keywords containing the line: sci-electronics/kicad **
Created attachment 306745 [details] results of emerge --info
All ebuilds appeared to run to normal termination with no elog email messages, but in each case the header on the Kicad widow read "KiCad (2011-nov-30)-stable after the merge.
Short story: this is not a bug. It's how upstream does things and I have decided I would stick to it so that users of other distributions do not get confused. Long story now. The way upstream does releases is not what is usually done, which is why the weird dates. They decide at some point that a particular bazaar revision is going to be their next release. They test it for a while and if they decide it's good then they do the release. However, the new version is going to be named after the date they do the release, not after the date the chosen bazaar revision was committed. See [1] for how the source tarballs (which we don't use for practical reasons) of the current release are named. And now this is where things get more complicated. The date you see in the window header or in the about box is another one. It comes from a file called version.txt which is included in the tarball or in the bazaar checkout. My guess is they only update it when they choose a given revision for a potential future release. That's why the live and unstable ebuilds show the same date. Confusing, I know, but that's the way things are. If it really bothers you, you may want to take it up to upstream and discuss it with them. Denis. [1] http://iut-tice.ujf-grenoble.fr/cao/
Thanks for the lucid explanation, Denis. This does seem a strange and confusing way to do things, but I can see why you have chosen not to monkey with it. The current version.txt on the site[1] shows the date 19 jan 2012 (in two formats). I would have expected the ebuild 20120119_p3256 to reflect the kicad sources of the same date, which upstream seems to regard as a release. In the light of your explanation, the 99999999 ebuild should then presumably show the same date. It looks as if the version.txt in the gentoo sources is somehow out of step.
