[12:02] mckinstry's still a bit active, although more on console-* than on d-i itself now === Kamion -> bed [12:04] fjp, smarenka, sferriol, zinoviev are the new people I don't know. That is, I've met fjp and he's friendly enough. No idea who the rest of the crowd is, though. [12:04] sleep tight [12:04] zinoviev's not around much - he's the partman god but kind of in a Judaeo-Christian way ;-) [12:04] heh, 'k [12:04] he lives in Bulgaria and AFAIK has net access once a month or something [12:05] partman is scary. [12:05] I wonder what'll replace it. Something scarier with pluggable components. [12:05] oh, btw, I checked with Kirsten and we're free on Sunday; fancy coming up to our local in North Cambridge for lunch? [12:05] that'd be very nice, absolutely. [12:05] cool [12:05] anyway, really bed now :) [12:05] :-) === DracosX [n=glenn@adsl-065-015-227-185.sip.rmo.bellsouth.net] has joined #ubuntu-boot === pitti [n=pitti@ubuntu/member/pitti] has joined #ubuntu-boot === zyga [n=zyga@2-mi2-1.acn.waw.pl] has joined #ubuntu-boot [10:09] WOO HOOO [10:09] pal is working on ppc now [10:40] infinity: so are we doing new udev today? pretty please? [10:41] I'm tryin', I'm tryin'... [10:41] Last round of tests for LRM.. I hope. [10:41] Unless I find glaring bugs, I'll upload this, and worry about the small bugs later. [10:53] hmm. since I have to switch d-i's initrd filesystem *anyway* since BenC helpfully modularised ext2 everywhere, it's tempting to just make the jump to initramfs [10:54] eh [10:54] hmm [10:54] initramfs++ [10:55] might as well break everything AT ONCE [10:55] \o/ [11:14] I'll upload my cdrom-checker thingy to make sure to break that as well, then. [11:16] hmmm [11:16] i have some mixed feeling about slamming swap on lvm [11:16] it's good for server, but bad for desktops [11:16] since it looks like the kernel can't resume from swap on lvm [11:16] but it's a kernel bug [11:17] what do you suggest guys? [11:17] This isn't going to be a default, is it? [11:18] no [11:18] it's a selectable option [11:18] Kamion: can i take a lock on partman-auto? [11:18] Then I don't see the problem in letting people select it. [11:18] well you don't get asked for: do you want partition foo on lvm? [11:18] fabbione: yes [11:18] If we have the capacity to warn them that swap-on-lvm means no suspend/resume, that might be nice. [11:18] you just select: "Trash my disk and install on LVM" [11:19] Oh, I see. It's all or nothing. [11:19] Hrm. [11:19] infinity: i was more thinking to do: if server install swap on lvm, otherwise swap on disk [11:19] Then maybe it's better to leave the swap ina real partition, unless we can fix the kernel bug (misfeature?) [11:19] Kamion: thanks. [11:20] but i am not sure if we can actually detect that at that stage of the installation === sivang [i=sivan@ubuntu/member/sivang] has left #ubuntu-boot [] === Keybuk [n=scott@descent.netsplit.com] has joined #ubuntu-boot [12:32] Keybuk : Is that the real you, or a floating IRC client? [12:35] real me [12:36] Cool. I'm dangerously close to declaring "enough testing is enough testing, damnit" and finally uploading LRM. [12:36] ok, I can upload my packages when you're done -- just ping me [12:37] Kamion: does busybox d-i now include nanosleep? [12:39] Mithrandir: note that there is no "nanosleep" binary now, I modified the klibc sleep binary to support 0.4 type arguments [12:40] Mithrandir: not AFAI [12:40] K [12:41] Keybuk: 'k [12:41] Kamion: can we do the same to busybox? [12:41] Mithrandir: feel free, it's in coreutils/sleep.c [12:41] thanks [12:41] might want to make it a config option [12:41] Alright, I declare nvidia on amd64 good, screw i386 (it should work if amd64 did), and now to double-check fglrx on i386 and upload. [12:41] Anything that breaks after that, I'll fix later. [12:42] we're using busybox or busybox-cvs those days? [12:42] busybox [12:42] So, another 30-60 mins, guys, then you can stop hating me. === Mithrandir ruffles infinity [12:53] grr, it uses doubles explicitly. Silly thing. [12:59] what's silly about that? [01:00] sorry, longs. [01:01] is strtotimespec a somewhat standard function? All the references I find to it are klibc refs. [01:04] no, it's just a klibc utility function [01:09] grr, handling stuff like 0.5m is harder than it seems. [01:20] yes [01:21] code reduciont++ [01:21] reduction even [01:22] Kamion: already removed 1.5K of duplicated code... this stuff rock hard [01:22] Kamion: i blame you for one thing only.. you didn't drag me into d-i earlier ;) [01:33] heh [01:33] fun, isn't it [01:34] Okay, rsync to chinstrap going. For better or worse, this is it. [01:34] When Australia's done fucking with my head (about 48 mins, according to rsync), I'll upload to jackass. === infinity goes to the store while he waits. [02:02] compiles now, at least \o/ [02:07] it seems to work as well [02:07] Kamion: want to see the diff or should I just upload it? [02:10] Mithrandir: just upload === Kamion is buried in partman.deb again. Quite fun though, just wrote basic code to merge multiple build-dependencies with differing relations/versions === fabbione kills more duplicate code [02:14] 8 -rw-r--r-- 1 fabbione fabbione 4542 2005-11-30 14:12 shared.sh === pitti [n=pitti@ubuntu/member/pitti] has joined #ubuntu-boot === infinity sheds a tear for his closed connection, and starts again, remembering --partial this time. [02:24] hhahah [02:25] ARGH GRRR HUPMEHIEIHUA#)(J [02:32] I'll make bootchart run in d-i if it sees "bootchart" on the kernel command line, sounds good? [02:32] so we can have it in the default images. [02:32] Sounds workable to me. === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-boot [02:32] Might be anice change to the real bootchart package too. I'm not sure if "I have it installed" should mean "I want to generate a chart on every single boot". [02:33] yup, it's a one-line change, so. [02:39] Keybuk: any comment on that? [02:39] Is that the keybuk who JUST uploaded bootchart? :) [02:40] (But not with that change) [02:40] yes [02:40] why would you have it installed if you didn't want bootcharting? [02:40] uhm, as in just a week ago, just just? [02:40] you might not want to chart every boot. [02:40] As in, just a few minutes ago. [02:40] it's annoyingly irritating to add a kernel command-line option to use something you have installed [02:40] blah [02:40] *shrug* just ignore the charts you didn't want [02:40] it doesn't overwrite them [02:40] Mithrandir: how big is bootchart-udeb? [02:41] it makes the boot slower. :-P [02:41] so? [02:41] only sliiightly slower [02:41] about 1.2s slower for me [02:41] Yeah, but I assume it slows down the boot process, which is a bit counter-intuitive, given why one installed it. :) [02:41] Oh, if it's only a second or two, whatever. [02:41] for bootchart-udeb it definitely makes sense for it to be optional [02:41] Indeed. [02:41] Kamion: -rw-r--r-- 1 tfheen tfheen 1796 2005-11-30 14:41 ../bootchart-udeb_0.9-0ubuntu3_all.udeb [02:42] ubuntu3 is already in the archive ;) [02:42] Keybuk: that's irrelevant. :-P [02:42] was an update to the right update-initramfs call === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-boot [02:42] (in this context) [02:43] Kamion: so "tiny". It's a couple of shell scripts. [02:48] did you make java udebs? :) [02:49] no. :-) [02:49] so you need to move the data out of the system to get it rendered. [02:50] so the only bit of the package that actually makes a significant impact on the boot isn't there anyway? [02:51] uhm, remember that it writes this to ramfs-es, and the primary use case is the live CD. [02:51] hence, it should not run by default, we use too much memory already. [02:52] I guess there's that [02:52] and boot times on live CDs can easily reach into a few minutes, so the hit will be even bigger. [02:53] Mithrandir: ok [02:53] Mithrandir: you want it in pkg-lists/base then? [02:53] after you fix this [02:53] Kamion: yes. === HiddenWolf [n=hc_brugm@212-127-236-81.cable.quicknet.nl] has joined #ubuntu-boot [02:55] Kamion: do you require a main inclusion report? [02:56] Mithrandir: no, not for new binaries [02:56] (from existing sources) [02:56] the bootchart source isn't in main [02:56] oh [02:56] yes, will do then [02:56] uggggh [02:56] Keybuk: ugggh? [02:56] is all of the java stuff I'm using in the real one in main? [02:56] it could drag in a lot if not [02:56] I think I'm using gcj, I cargo-culted most of it though [02:57] rsvg at least is in universe [02:57] Keybuk: you're only using java-gcj-compat, which is in main [02:57] also do you provide any script for users to run to turn a bootchart.tgz into a png file? [02:57] Keybuk: no. [02:58] you should probably do that [02:58] librsvg2-bin is in universe, but librsvg2 is in main, so that's a no-brainer to promote [02:58] ok [02:58] not to worry then [02:58] it's probably an error that it isn't in main already, in fact [02:59] Kamion : Still don't have fancy "follow a library to find its debug, doc, and bin stuff" support? [02:59] Kamion: pcmciautils change for you ... dep on (module-init-tools >= 3.2.1-0ubuntu1) and always call "modprobe -ba modules..." [03:00] that way if people put yenta_socket, etc. in the blacklist, it won't get loaded [03:04] Kamion: ok, i am done with partman-auto for today. ubuntu2 is up === infinity gets out and pushes rsync. [03:11] ok, inclusion report written, now I just need to wait until the new bootchart appears in the archive, so I can merge my changes to it. [03:13] Keybuk: do you have your ubuntu3 somewhere? I'm bored of waiting. [03:15] It'll be on archive.u.c in about 3 minutes. [03:15] It's on jackass, the trigger just needs to run. [03:16] I think it was just a three character change, though. s/ -t// in the postinst. [03:16] There, mirrored: http://archive.ubuntu.com/ubuntu/pool/universe/b/bootchart/ [03:17] infinity: not yet [03:17] thanks [03:19] Keybuk: is that liable to hit Debian? (so that I know whether I need to create an Ubuntu branch) [03:19] actually, I should probably have an Ubuntu branch anyway [03:19] though, [03:19] make: *** No rule to make target `patch-stamp', needed by `build'. Stop. [03:20] Kamion : Feel like doing pre-emptive overrides for lrm-2.6.15? [03:21] Keybuk: do I need -a to make -b work? I don't actually want wildcarding here [03:22] infinity: is elmo not around? I hate munging overrides when I don't have to :) === infinity pings elmo. [03:23] Kamion: so we only miss the last bit.. that's going to be the most annoying one.. [03:23] infinity: -nvidia-legacy gone away? [03:24] Kamion: pvremove -ffy on the right device :) [03:24] fabbione: shared.sh is a pretty sucky name, since I'm guessing you have that in /lib/partman/ [03:24] please rename that to auto-shared.sh or something less namespace-polluting :) [03:24] Kamion: yes... /lib/partman/ [03:25] ahhh crap [03:25] i forgot to add the note.. name was clearly up for discussion :/ [03:25] it's really missing from the changelog :) [03:25] sure [03:25] will do [03:25] Kamion : Yup. [03:26] Kamion : As well as the i386 SMP variants (obviously) [03:27] ok, well if elmo doesn't ping back soon, let me know [03:27] Kamion : And elmo seems decidedly out. [03:27] hm, ok [03:27] 13.5 hours idle with an away message is never a good sign. [03:29] infinity: err ... well somebody's at home because lrm just got NEWed [03:29] although only the source [03:29] Kamion: done :) [03:30] fabbione: ta [03:30] no problem [03:32] Kamion : Oo. [03:32] Kamion : He's being stealthy, then. [03:33] * Stop using crazy *_minor versioning scheme, and just have the oddball [03:33] packages use a scheme of UpstreamVer+SourceVer (ie: 1.0.7676+2.6.15-1) [03:33] woo GO INFINITY [03:33] Keybuk : Congratulations on uploading a (supposedly) 3 character change and making your package FTBFS. :) [03:33] ok, I think I'll just do the binaries anyway [03:34] infinity: it should have FTBFS-ed before too. [03:34] infinity: I'll handle it. [03:34] Keybuk: ^^ [03:34] Ahh, it wasn't a change in the current upload, then? [03:34] nope [03:34] I'll stop poking fun, then. :) [03:34] I think he broke it in 0ubuntu1 or something, though. [03:36] infinity: ok, binaries should sail through the queue now with any luck [03:37] \o/ [03:38] I'm really hoping that, modulo one or two very quick bugs, LRM manages to maintain itself for the rest of the release cycle. [03:39] It's in better shape now, at any rate. [03:39] And it's a single-edit operation to do ABI bumps. [03:39] I wonder if it'll build on the arches I didn't test... === infinity is suddenly paranoid that he didn't bother testing on the PPC machine here. [03:41] infinity: YTL anyway [03:44] Kamion : Do you know if elmo managed to NEW it before cron.daily triggered? [03:44] Oh, wanna-build is running right now, guess I'll find out in 2 minutes. [03:45] Yup, it made it. [03:48] bootchart uploaded, \o/ [03:49] actually, can we have the bootchart package in the live seed? [03:50] it'll stop bootchart and generate the graph at the end of booting. [03:51] sure, if it's (a) done conditionally by a boot arg and (b) not too big [03:52] it's a noop if bootchart isn't running. [03:52] the .deb is 97k [03:53] ok [03:53] Damnit, it doesn't build on powerpc. I jinxed myself. [03:53] oh, bigness including dependencies [03:53] make sure we already have all the java crap in there ... if not then it's more difficult [03:54] Kamion: ok, will check [04:02] bah, unstable != dapper [04:02] dput should refuse to upload unstable packages to ubuntu === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-boot === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-boot [04:13] Alright, build failure on powerpc is the kernel's fault. [04:13] New linux-meta is blocking on linux-image -6.8 now. [04:13] Keybuk : As before, I'll upload udev and linux-meta at the same time, once this is all sorted. [04:15] Kamion: current live still seems busted from my rootskel breakage? [04:16] uhm [04:16] and it thinks Norwegian should default to Macedonian keyboard layout. [04:16] That seems sensible. [04:16] localechooser is on crack [04:16] based on my selection of keys, it thinks de-latin1-nodeadkeys is a sensible choice. [04:17] heh [04:17] there's an off-by-one-error somewhere [04:17] when I select Norwegian from the list, it displays macedonian [04:17] but it actually is norwegian, since I can type [04:17] Kamion: any idea what's up with that? === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-boot [04:19] not offhand, try DEBCONF_DEBUG=20 [04:20] current live is busted, but I think not from your rootskel breakage? it's waiting on new udev to work properly [04:20] that too [04:20] you can bring it up if you start up udev by hand [04:20] (I'm running it over --> there right now) [04:21] I get raw debconf stuff at my terminal if I don't pass MENU to isolinux [04:21] looks like localechooser breakage. [04:21] since it substs in the wrong value. [04:22] keymap_ask: trans: no-latin1 [...] SET kbd-chooser/method Makedonsk - mk [04:22] I guess it's an off-by-one error [04:25] that would be kbd-chooser not localechooser surely [04:25] probably a missing comma in some debconf template translation [04:25] what locale? [04:26] sorry, additional comma I mean [04:26] nb [04:26] msgstr "Tsjekkisk, cz-lat2" [04:26] ideally that comma would be escaped ... [04:27] (in console-data) [04:27] hmm, it is escaped [04:27] also T comes after M [04:27] and N [04:28] the list is sorted according to the msgid, not the msgstr; Czech < Macedonian [04:28] ah [04:29] I think that's it, but I want to know why the comma-escaping isn't working, since there are *lots* of strings like that for various languages in console-data; rather than trying to patch all 133 of them I'd prefer a real solution :) [04:33] it is escaped in the input to cdebconf. [04:37] kbd-chooser.c doesn't seem to handle escaping of , at all [04:37] it just uses strchr [04:37] are you sure it's supported? :-) [04:37] but it doesn't search for , ... [04:38] lim1 = strchr (p1, ','); [04:38] lim2 = strchr (p2, ','); [04:38] before line 571 [04:38] SMURFIX [04:38] that's an Ubuntu patch [04:38] heh [04:39] fix away if you've got a handle on it; probably wants to walk along the string skipping \ and *then* checking for , [04:39] why isn't that put in upstream? [04:39] it's all munged up with the keymapper stuff I think [04:39] gnrrr [04:40] I need to go to a shop and fetch some tickets for London, but I'll look at it when I get home. [04:40] it's part of kbd-chooser/method which is an Ubuntu patch [04:40] I know where the bug is, at least. [04:40] the delta towards debian in the installer on those kinds of things suck. :-( [04:41] uh-huh, tell me about it [04:42] let's revert * to the original d-i :) [04:42] speaking of which, had any time to look at the custom widgets stuff? [04:42] let's keep only choose-mirror adn we are done [04:44] Mithrandir: no :-/ [04:45] I might have time before going to London, I don't know yet. [04:45] I promised to do an ia32-libs update for bdale tonight, so. === BenC [n=bcollins@debian/developer/bcollins] has joined #ubuntu-boot [04:54] Kamion: ping [04:56] BenC: pong [04:56] what is it that needs to be updated for the firmware path changes, hotplug or udev? [04:56] hotplug is dead, so udev [04:56] but udev already has been AFAIK ... [04:57] or at least will be in Scott's new udev [04:57] No, it hasn't been uploaded. [04:57] So, firmware is broken until that happens. [04:57] But we can't upload udev nutil linux-meta is ready to change. [04:57] it needs to be within 24 hours, so I can complete the linux-meta update [04:57] Or the world will go tits up. [04:57] can we upload them at the same time? [04:57] BenC : I'm doing linux-meta at the same time as I do udev (which is right after LRM builds everywhere) [04:58] I don't want linux-meta updated until udev is, and vice versa, so this needs to be synced [04:58] ah, I already had a linux-meta done, since it has changes for {686,k7}-smp targets [04:58] So, get me a working -6.8 by the time I get up tomorrow, and you'll have it all working when you wake up. :) [04:58] just pending upload [04:58] Yeah, I made those same changes here. [04:58] Oh well. :) [04:58] ok, then it'll be in your hands [04:59] what about udev though...we need that for linux-meta [04:59] BenC : I'm doing linux-meta at the same time as I do udev ... === Kamion watches the conversation going round in pretty little circles [04:59] doh [04:59] lol === BenC wipes his lcd === infinity goes to bed to dream of fixed kernel headers. :) [05:00] infinity: so what about this udev thing? :) [05:00] ok, so you'll upload lrm, linux-meta and udev in the morning [05:00] I'll have -6.8 done in a few hours (ahead of time actually) [05:01] "it appears to be a white hole" [05:02] just wanted to be clear on things before I sent the email to u-d-a [05:02] figure we should warn people before totally destroying their systems [05:13] Fixed kernel headers? [05:13] Is lkh broken? [05:13] Or real kernel headers? [05:26] meh, lost my IRC window [05:26] silly "hide all windows" button [05:27] Kamion: I've sent the modprobe change to Rusty directly, so it may hit Debian via upstream -- Marco also may pick it up. it's definitely required in our packages though, we actively use the blacklist and encourage people to use it -- and have open bugs already about things like pcmcia ignoring it [05:28] Kamion: the -ba thing is just if you expand a list of modules, -a means "more than one module on the command-line" ... if you just put one argument to modprobe that may be an alias, don't use -a [05:28] infinity: why did bootchart FTBFS? [05:29] infinity: uh, don't upload the udev I gave you on Monday, is verrry broken :) [05:31] Keybuk: ok, change made locally, will upload in a bit [05:33] typical usage is something like: [05:33] SUBSYSTEM=="input", PROGRAM="/sbin/grepmap --udev", RUN+="/sbin/modprobe -ba $result" [05:33] where $result is a space-seperated list of the modules grepmap output [05:34] also things like: [05:34] SUBSYSTEM=="scsi_device", SYSFS{type}=="1", RUN+="/sbin/modprobe -b st" [05:34] which loads the st module on tape devices, but lets the user blacklist st if it panics their kernel or something [05:35] in this case we only ever load one PCMCIA bridge at a time, so just -b is fine [05:35] yup [05:54] Keybuk: you hadn't removed the dependency on patch-stamp from the build target [05:58] heh, how did 0ubuntu1 build then? :) [06:07] nfi [06:07] 0ubuntu2 never built for me. [06:08] I bet I included patch-stamp in the diff.gz or something silly ;) [06:10] Is patch willing to create empty files? [06:10] yes [06:10] at least I think so [06:10] I thought it wasn't. [06:11] : tfheen@xoog /tmp > mkdir a b [06:11] : tfheen@xoog /tmp > touch a/file [06:11] : tfheen@xoog /tmp > diff -Nru a b [06:11] : tfheen@xoog /tmp > [06:11] doesn't look like it. === HiddenWolf [n=hc_brugm@212-127-236-81.cable.quicknet.nl] has joined #ubuntu-boot [08:02] Kamion: ping [08:03] Keybuk: pong [08:04] Kamion: could you drop hotplug from ubuntu-meta and upload nowish [08:04] uh, ok, give me ~10 minutes [08:06] ok [08:06] (have to merge everywhere etc.) [08:06] yah [08:07] I doubt udev will hit this cron.daily [08:07] er ... we don't have new linux-meta yet? [08:07] will this break the world? [08:07] correct [08:07] some bits of it [08:07] mdz exercised his "DOIT" option [08:08] ok [08:08] (I'm updating linux-meta now) [08:09] it'll break lrm users, and adam will need to clean the coffee from his monitor and keyboard in the morning but otherwise should be ok [08:10] breaking lrm will break the installer too of course [08:10] but the installer will probably be broken anyway for a bit [08:25] Keybuk: ubuntu-meta uploaded [08:30] I'm keeping grepmap-udeb, right? [08:30] right [08:31] of course, I could always PUT THE MODPROBE RULES INTO THE UDEB [08:31] la la la [08:31] that would be nice ;-) [08:31] d-i ready-ish to upload [09:35] so should the installer be using udevplug -W after installing new modules on the system? [09:35] (i.e. is -W the right option to use?) [09:35] it can do, certainly [09:35] where "installing new modules" == "modprobe calls" [09:36] no, as in udpkg -i foo-modules.udeb [09:36] right [09:36] well, and then modprobe calls, yeah [09:36] you could do: udevplug -f [09:36] not documented [09:36] if the original modprobes failed [09:36] -F? [09:36] sorry [09:36] -F [09:36] it's not really to catch failure, it's to wait for things to settle [09:36] but the original modprobes may not have failed if the alias did cause some generic device to be loaded [09:37] just adding modules to the filesystem won't do anything [09:37]