On Sun, Nov 15, 2009 at 7:35 PM, Robin H. Johnson <robb...@gentoo.org> wrote: > On Sun, Nov 15, 2009 at 09:25:45AM +0200, Petteri Räty wrote: >> Doug Goldstein wrote: >> > GLEP 27 [1] seems pretty stagnant and I'm planning on giving it a bit >> > of a refresh and actually implementing it. Now before I do this I'm >> > not in love with the format in tree but I haven't decided on a format >> > exactly in my head. So that being said, I'm sending this out looking >> > for some opinions or ideas for my new GLEP. One of the obvious things >> > I'll cover is all the ambiguity of the GLEP with regard to the data >> > inside each of the files. >> > >> > [1] http://www.gentoo.org/proj/en/glep/glep-0027.html >> One idea worth considering is making users just ebuilds with a >> supporting eclass. > While I'm hugely in favour of having consistent UID/GIDs with no > conflicts over all Gentoo machines, I feel one of the reasons that the > GLEP failed was that users required by ebuilds changed over ebuild > versions, and the GLEP didn't seem to handle that well. > > Cases I've seen in the tree: > - username change (slocate -> locate) > - homedir change > - shell change >
Which would seem to mean that Petteri's suggestion would work better since that would allow us to version/upgrade user/group data. -- Doug Goldstein
