=== ripps_ is now known as ripps === mzz_ is now known as mzz [19:00] hyperair, I saw your new mesa flag in git. which changes are needed in the packaging? I can add it to x-updates [19:01] tormod: add --with-dri-searchdir=/usr/lib/dri:/usr/lib/dri$(DEB_HOST_GNU_TYPE) to configure [19:02] er wait [19:02] tormod: add --with-dri-searchdir=/usr/lib/dri:/usr/lib/dri/$(DEB_HOST_GNU_TYPE) to configure [19:02] i'll come up with a debdiff for mesa [19:02] hyperair, ok will do [19:03] tormod: by the way, it seems that one of the patches in mesa hits on symlinks and comes up with issues when quilt popping [19:03] hyperair, yes if you run quilt on your git tree it fails [19:04] tormod: no, i'm talking about an apt-get source'd directory [19:04] if you unpack an orig.tar.gz the links have been dereferenced [19:04] oh still? [19:04] yes [19:04] and sorry i got the config flag wrong [19:04] it's --with-dri-searchpath [19:06] hyperair, which mesa patch? [19:07] tormod: 35506dec6734a1ae777dc16981cb19e514f41897 [19:08] oh I see I was thinking the other way, a Ubuntu patch on git. but you have git patch on orig.tar.gz... [19:10] that's your git patch, I don't see where it hits any symlink? [19:11] tormod: oh sorry. you meant the quilt issue? [19:11] yes [19:11] it's the one that hits on the radeon symlinks [19:11] r{2,3,6}00/radeon_dma.c [19:11] they all point to "../radeon/radeon_dma.c" [19:11] (I told you had seen your mesa patch in git :) ) [19:11] yes I know [19:12] but that's fine if you're using orig.tar.gz [19:12] my bad. my text-to-speech converter in my brain isn't working well [19:12] and yes i'm using orig.tar.gz [19:12] the problem is when you quilt push -a to work on a patch and then quilt pop -a [19:12] the next debuild fails because of unrepresentable errors [19:13] which patch was that, I can not find it (although I remember spending hours on it= [19:13] but there are no symlinks in the unpacked orig.tar.gz right? [19:15] there are. [19:15] at least, you can't put links inside diffs, and it was already there after i apt-get source'd it [19:15] couldn't you just quilt remove the said files? [19:15] it'd be so much simpler =\ [19:28] oh yes you are right, there are links [19:29] you talking about 109_revert-dma-reuse.patch ? [19:29] it's gone upstream so I don't use it [19:32] I think quilt is not symlink aware, so if you use quilt to make the patch, it adds hunks for each link + original [19:32] but it works on the builder, so there is something I don't get [19:33] are packages unpacked differently on the builder vs apt-get source? [19:37] tormod: quilt can push -a once, and pop -a once. subsequent attempts will afil. [19:37] fail* [19:38] since it only pushes all the way once for building, it works fine [19:38] also, quilt tracks each change by file. [19:38] i.e. only if you've quilt add-ed before you edited your file [19:39] symlinks shouldn't make a difference unless you quilt add-ed your symlink, (which is what appears to have gone wrong here) [19:40] tormod: i'd like to get this fix into karmic as a SRU. i'd put karmic-proposed in the changelog, right? [19:41] yes [22:56] wgrant: re your question earlier on #u-devel, the debian archive already started accepting v3 source packages === jbarnes_ is now known as jbarnes [22:58] jcristau: I know. Just wondering when it will actually be allowed. [23:02] http://lists.debian.org/debian-devel-changes/2009/10/msg02608.html [23:02] so, now. [23:53] jcristau: ... fuck.