/srv/irclogs.ubuntu.com/2012/05/28/#ubuntu-x.txt

mlankhorstRAOF: yeah needs testing though06:37
mlankhorstplymouth was really annoying, had to rebuild that one first without shlib depends before I could install renamed libdrm06:41
RAOFYeah, essential: yes packages make the baby Jesus cry.06:43
mlankhorstRAOF: I had to disable the binary driver packages because I wasn't sure it would be needed to rename those. Newer versions will probably work with new abi too.06:45
tjaaltonright06:45
tjaaltonalthough, probably the 'stable' version needs to be renamed06:45
RAOFThat's a fair point; we'd need to ship {nvidia-current,fglrx}-updates, though.06:45
mlankhorstindeed06:46
mlankhorstbut since this was meant to  be a proof of concept for testing mechanics I felt just using open source drivers was enough for now :)06:47
tjaaltonsure06:47
mlankhorstthe part that scares me is the versioned replaces, I could install xfonts-base on top of the renamed package because it was newer06:48
tjaaltonwhy would you need a renamed xfonts-base?06:49
mlankhorsttjaalton: this is just testing mechanics06:49
mlankhorstso why not06:49
tjaaltonguess there are plenty of packages to test with already :)06:49
mlankhorstxfonts-base was special because of its dependency on xfonts-utils06:50
tjaaltonok06:50
mlankhorstI learned from that renaming xfonts-utils is non-trivial and an update would probably be better in that case06:51
tjaaltonlikely no need for either06:52
mlankhorsttrue, but good to test regardless06:52
mlankhorstand maybe not for quantal, but there are some other stacks to test too :)06:53
tjaaltoncore fonts, who uses them anymore anyway :)06:53
tjaaltonno great new features to be seen there06:54
RAOFAre you sure we shouldn't backport xfs? :P06:54
tjaaltoni'm sure we should drop it :)06:54
mlankhorsttjaalton: yeah but if xshiny-new-bright-fonts comse along that requires a new xfonts-utils06:54
tjaaltonwell maybe not drop it, low maintenance06:55
RAOFYou mean: drop it as soon as someone files a bug against it. ?06:56
tjaaltonhehe06:56
mlankhorstwell duh D:06:56
mlankhorstdont drop it, just replace by empty package, see who complains, bahah06:56
mlankhorstlets see if the renamed stack installs on the cd06:57
mlankhorstim not going to let it anywhere near my dummy pc's hard drive06:58
mlankhorsthm console-setup gets annoyed too07:15
mlankhorstI guess now the goal is to get it installed without causing ubuntu-desktop and ubuntu-minimal to be removed :)07:26
RAOF:)07:31
tjaaltonthat would mean the provides are broken07:33
mlankhorsttjaalton: yeah can't do versioned provides07:34
tjaaltondo they actually need those?07:35
mlankhorstthat's the reason they're broken..07:35
tjaaltonhuh07:35
RAOFWhat's depending on explicit versions of the stack?07:38
mlankhorstusually because of a Depends ${shlib:Depends}07:47
mlankhorstwhich means basically anything that links to any library in our stack07:48
RAOFOh, right.08:14
RAOFI thought we weren't updating libraries, though.08:14
tjaaltonlibdrm..08:14
tjaaltonat least08:14
tjaaltonprobably not due to that though08:15
RAOFRight. But there's nothing outside the stack that depends on libdrm (?).08:15
tjaaltonyeah08:15
mlankhorstRAOF: yeah but plymouth/intel-gpu-tools have versioned depends08:17
mlankhorstand console-setup had a versioned depends on xkb or something08:17
RAOFYeah, but I was counting those as a part of the stack.08:17
RAOFAh, console-setup?  Hmm.08:17
mlankhorstatm im just trying to get to 0 removes, so by pushing new packages im keeping track of which packages break08:18
tjaaltonconsole-setup could drop the versioned depend08:18
tjaalton>= 0.9..08:18
mlankhorstthats what my rebuild does08:18
tjaaltonwe're at 2.5 now08:19
mlankhorstget it in for quantal, then :D08:19
mlankhorsti dont have the rights to push anything08:19
tjaaltonit should go as an SRU for precise as well08:20
tjaaltonpreferably before 12.04.1..08:20
mlankhorsttjaalton: we're not sure yet what packages we would need, this is just a test of the mechanics08:26
tjaaltonso it's not blocking anything?08:27
tjaaltoni'm confused08:27
mlankhorsterm, it depends if we decide we need a new xkeyboard-config and x11-xkb-utils or not08:27
tjaaltonit's a nice-to-have, not something hw enablement needs08:27
mlankhorstif you want it drop version in console-setup and sru it, if not just dont ship those 2 renamed :-)08:29
tjaaltondropping the version is easier08:30
mlankhorstoh right, need to rebuild the xorg package too08:31
mlankhorstRAOF: any objections if the xorg source package will be treated specially? makes things a lot easier08:32
RAOFNone at all.08:36
mlankhorstperfect :-)08:36
tjaaltonyeah, big thumbs up from me08:36
tjaaltonmost of the metapackages it builds make no sense08:37
tjaaltonfor the backports08:37
mlankhorsttjaalton: those metapackages need to be tweaked to depend on the oldname | newname08:37
tjaaltonright08:37
mlankhorsthm, how can I see why apt-get install .*lts-backport-quantal decides to remove some packages?08:59
RAOFmlankhorst: I find that aptitude's resolver is sometimes more usefully verbose.09:01
mlankhorstyeah but it doesn't allow regexp install09:02
RAOF~nlts-backport-quantal09:02
mlankhorstty :)09:02
RAOFYou can also play around with the debug options (pass -o to apt-get, check out "man apt.conf" for the things you can actually *pass* to apt-get)09:03
mlankhorstoh libxklavier16 was going to be removed it seems09:05
mlankhorstxkb-data >= 0.8, sigh09:05
tjaaltonuh09:07
mlankhorstklavier is flemish for keyboard :)09:09
mlankhorstoh fun, it keeps adding the versioned depend on xkb-data back09:19
tjaaltoncheck if there's a control.in09:19
tjaaltonsince it's a gnome package..09:20
mlankhorstyeah it has ;s09:20
mlankhorstjust independently noticed it 09:20
tjaaltonso modify that instead09:20
mlankhorstyeah i did, i keep running into weird new ways of packaging :-)09:21
mlankhorstalmost all packages seem to have their own silly way to break09:23
mlankhorstsuccess! It only replaces the X stack itself now09:25
tjaaltonnice09:27
mlankhorststill some package failure it seems09:29
mlankhorstoh right.. cant install libgl1-mesa-swx11 at the same time as the normal version09:29
* RAOF wonders why we even build that package09:33
mlankhorstno idea :-)09:36
mlankhorstprobably carried over from debian and removing was more work09:37
tjaaltonjust that it's become somewhat obsolete on the hw we support09:38
tjaaltondue to llvmpipe09:38
tjaaltoni'd like to drop the 8 & 16bit osmesa packages..09:39
mlankhorstoh woops, probably broken due to rename09:39
mlankhorstlooks like mesa packaging needs more love09:41
mlankhorsthm, how do the bug scripts work?09:56
mlankhorstseems they're path dependent on package name, but with renamed package it looks like it would break09:57
tjaaltondo we want bugs for them?-)10:21
tjaaltonbesides, you can just use the original names/links10:21
tjaaltonwe don't really want bugreports against the renamed source packages..10:21
mlankhorsti have no idea how the mechanism is supposed to work for that10:22
mlankhorstbut ok I'll trust you on that one. :s10:24
tjaaltonwell, it's just pointing to the xorg bug script10:25
tjaaltonbut not sure if apport will use the package source name anyway10:26
mlankhorstill just keep it broken for now10:28
mlankhorstI don't think it's a source package, but the binary package being used10:29
tjaaltonhm, apport isn't that useful.. my gpu hung and apport offered to report a bug, but after clicking 'continue' the window just closed and didn't open lp for finishing the submission12:58
mlankhorstoh seems we DO need all open source drivers including the ugly ones for testing, else xserver-xorg-video-all fails12:58
tjaaltonor you create xserver-xorg-{input,video}-some12:59
tjaaltonprobably not a good idea..12:59
mlankhorstwell i can replace the X stack now at least13:02
mlankhorstI think the upgrade is breaking because of the last missing ones to satisfy the xserver-xorg-video-all install13:03
mlankhorstmy goal for today is simply being able to install the renamed stack without removing any packages that haven't been renamed.13:08
tjaaltonhmm, does kernel 3.4 require something special for intel? it's not using intel kms here..13:35
tjaaltonbut vesa, which probably makes the kernel hang hard after maybe 10s13:35
mlankhorstanything in xorg log?13:37
tjaaltonjust that it falls back using vesa13:38
tjaaltonin fact it doesn't initialize drm at all13:38
tjaalton..13:38
mlankhorsto.o13:38
tjaaltonrevealed by dmesg13:38
mlankhorstwell there you go, then13:39
mlankhorstthe pae kernel should work on non-pae hardware right?13:40
tjaaltonnot sure13:42
tjaaltonactually no, it doesn't work13:43
tjaalton3.4 works on an old i965 based machine13:47
bjsnideris there still a non-pae i386 kernel?14:04
tjaaltonit's getting removed from quantal14:06
tjaaltonor not built anymore, dunno14:06
tjaaltonthe problem I had was missing dri modules from the drm-next build14:12
tjaalton..14:12
mlankhorsthm during upgrading it removes ubuntu-desktop, but i can install it again afterwards, sigh14:15
mlankhorstI guess I'll need to specify ALL the packages to apt-get in 1 go :)14:21
tjaaltoncheck with aptitude why it removes u-d14:21
mlankhorsti know why, i missed some packages causing xorg to be removed14:22
mlankhorstso I'm just going to create a dummy package in xorg that has a dependency on the whole renamed stack14:23
tjaaltonok14:24
mlankhorsthm, how do I sed variables in a makefile?14:30
mlankhorstnm I think ill split it off to its own package14:31
mlankhorstbryceh/raof: I can install the updated stack from q-lts-backport now, see the mail I sent on the ubuntu-x mailing list :-)16:46
brycehmlankhorst, sweet23:36

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!