/srv/irclogs.ubuntu.com/2012/05/30/#ubuntu-installer.txt

twbWhat is the status of live-initramfs/-config/-build as at precise?04:15
twbSpecifically, I have a headless x86 router that I wish to PXE boot off a prepared SOE.  I have had good experiences in the past building such with live-initramfs & debian testing -- it seems a lot better than casper/lucid.  Since I don't want to deploy testing's moving-target in production, I'm looking at live-initramfs on precise.04:16
twbBack in lucid days, live-initramfs on ubuntu was in the "nobody has tried it before, good luck!" territory, but I get the impression it is substantially better as at precise04:17
twb(BTW if there is a better channel to discuss this, please redirect me there.)04:17
cjwatsontwb: I still can't claim any particular experience with live-initramfs.  I'd like to rebase our casper package as some patches on top of live-initramfs at some point, but have never had the time.  You're welcome to try.08:11
cjwatsontwb: We use live-build nowadays, as of oneiric.08:11
twbyou use live-build + casper?08:13
cjwatsonYes08:14
cjwatsonWe still use livecd-rootfs, but really just as a place to put our live-build configuration these days.08:14
twb$coworker got oneiric working with live-build + live-initramfs08:14
twbI don't do non-LTS, so I'm going to see how far I get with twb-bootstrap + live-initramfs/live-config + precise08:15
cjwatsonI'm sure live-initramfs isn't desperately hard; random patches here and there.  It's not that dissimilar to casper.  The work is just in going through with a fine tooth-comb and making sure nothing has gone missing.  Probably some days' tedious work if we wanted to migrate Ubuntu proper.08:15
cjwatsonThere'll certainly be miscellaneous hacks that aren't in live-initramfs.08:15
twbIn my case it's more "does it work in twb's unusual use cases"08:16
DeprofundisHi08:21
Deprofundisi'm trying to modify ubiquity and i want to know if there are any rule that i've te respect to do that ?08:23
cjwatsonHow do you mean?08:24
twbDeprofundis: are you modifying it for in-house use, or are you going to upload the result into Ubuntu?08:24
cjwatsontwb: Deprofundis doesn't have Ubuntu upload access08:24
twbcjwatson: well he might be sponsored or something08:24
twbBut obviously if it's in-house he can do what he likes08:25
cjwatsonAny ubiquity change would require review by us08:25
Deprofundisi would like to upload the result ... the project is about integration of  OneConf in ubiquity !!08:25
cjwatsons/upload/merge/08:25
cjwatsonUm, if you're adding new UI, you need to get design review first (and personally I'm extremely reticent to approve the addition of new pages to ubiquity - we have a mandate to keep the installer as simple as we can)08:26
twbWhat's a oneconf?08:26
Deprofundisi've had talk about this with mpt08:27
cm-ttwd: oneconf sync the list of your installed software08:27
cjwatsonTry extremely hard to fit into prevailing coding style.  Try to write things in plugins rather than in core code if you can manage it.  Get code review early and often.08:28
cm-ttwb: we try add a step "install from sync"08:28
Deprofundisand i already have the page ... anyway i will developped : adding the page required ( withou making merge ) and you can review and approve it if you want !!08:28
cjwatsonI'm still deeply unconvinced about adding another page that everyone has to see.08:28
twbcjwatson: +108:28
cm-tnot really08:28
twbcm-t: so it what, uploads your dpkg --get-selections to a configuration management server on the LAN?08:29
cm-tI think we should had introduce ourself better08:29
cm-t^^08:30
twbcm-t: if you have a project page just link to it, i'll go read that08:30
cm-thttps://docs.google.com/a/canonical.com/document/edit?id=1bZ4yQIVgGaUGSYu3qiUHnQt3ieBZoqunP_DcleHCr3I&pli=1#sync08:30
cm-tat the "install from sync" step08:30
twbSorry, docs.google.com doesn't work in w3m08:30
cm-tok08:30
cm-tso to be fast:08:30
cjwatsonAh, if it's a different option alongside Try/Install, that's not so bad08:31
twbHaha, that link doesn't work in midori either08:31
cjwatsonI'm somewhat worried about the potential increased bug count though.  apt-clone restores have been bad enough for that.08:32
cjwatsonNeeds excellent automated tests.08:32
cm-ton the welcome screen you have an option "install from sync"  if you select that you have the step connect network then ubuntu-sso  then oneconf (select what to clone)  then normal install + adding package08:32
Deprofundisyes i know ... but i don't know if we can invest in automated tests ...08:33
cjwatsonYou must08:33
twb...and then google docs / midori crashed my tegra2 GPU driver...08:34
* twb make a note not to test google docs URLs in X anymore08:35
cm-ttwb: https://docs.google.com/a/canonical.com/document/export?format=pdf&id=1bZ4yQIVgGaUGSYu3qiUHnQt3ieBZoqunP_DcleHCr3I  ← suppose to be the pdf export08:36
Deprofundisis there any topic about modification ubiquity , where i can find how to start modify ubiquity08:37
twbcm-t: thanks08:37
cm-ttwb: np08:37
cjwatsonDeprofundis: https://wiki.ubuntu.com/Installer/Development08:40
cjwatsonhttps://wiki.ubuntu.com/Ubiquity08:41
cjwatsonThere's no "how to be a developer" document as such08:41
cjwatsonYou already need to be a fairly experienced Python developer08:41
cjwatsonAnd familiarity with POSIX shell is very helpful too08:41
antarusI would argue the opposite, it is fairly easy to screw yourself in shell, a bit harder in python ;)08:42
twbhttp://www.opengroup.org/onlinepubs/9699919799/ and #bash can help with the latter08:42
NCommand1rcjwatson: what is the bes way to create precise-proposed branches for d-i/base-installer? I need to prepare SRUs for highbank enablement08:43
xnoxNCommand1r: you cannot create -proposed branch if it doesn't exist already.08:44
cjwatsonantarus: But in the stuff these guys want to do, there's much more Python involved08:44
* antarus nodes08:44
cjwatsonxnox: not applicable here, base-installer doesn't use the UDD branch scheme08:45
antaruswe've only done custom oem-config stuff08:45
NCommand1rxnox: we had ~ubuntu-core-dev/debian-installer/distro-proposed08:45
cjwatsonNCommand1r: lp:~mcasadevall/base-installer/precise-proposed08:45
cjwatsonor ubuntu-core-dev if you lilke08:45
=== NCommand1r is now known as NCommander
cm-tcjwatson: expect for the lxc or chroot for installing all package from sync08:45
* xnox sorry =)08:45
cm-texept*08:45
NCommandercjwatson: thanks, will have a bug with proposed debdiffs up soonish08:46
twbGrr, why isn't ubuntu-keying package in Debian :-/08:54
twbhttp://paste.debian.net/171882/ why is initramfs-tools having a sad during cdebootstrap of precise?09:04
twbNever mind, I'll bitch at #ubuntu-server instead09:05
cjwatsonDon't waste your time with cdebootstrap; use debootstrap instead.09:07
twbI'm sure there was a reason I was using it before09:08
cjwatsonThere was a point to cdebootstrap about eight years ago.09:08
twbBut yeah, that's what I'm doing now09:08
cjwatsonBefore debootstrap gained its own dependency resolution engine.09:08
twbMaybe just inertia09:08
twbcjwatson: hm, really?09:08
cjwatsonSince then it's been a waste of effort IMO.09:09
cjwatsonAnd we've never put any effort at all into making cdebootstrap work with Ubuntu.09:09
twbLast time I looked all three (^ + multistrap) just went and tried to configure things at random until there was nothing left to configure09:09
cjwatsonNo, it's not that simple.09:09
twbIt was when I looked at the code...09:09
cjwatsonAnd it's about selecting packages to install, not about what order things are done in.09:09
twbOh, you mean which to dl, before it unpacks them?09:09
cjwatsonYes.09:10
twbOh well.  I don't normally have any problems with tha09:10
cjwatsonOnce upon a time that was much more hardcoded in debootstrap than it is now.  But that was fixed in breezy or something.09:10
twbThe paste above is because mawk's alternative hasn't been set up before initramfs-tools is configured, I think09:10
cjwatson*shrug* I'm not going to work on cdebootstrap.09:10
cjwatsonSorry :-)09:10
twbBecause I never trusted *strap's resolution, I *always* told all of them to only pull down apt + its deps, and then did the rest in the chroot...09:11
cjwatsondebootstrap --variant=minbase09:11
twbYeah that09:11
cjwatsonIf you want that09:11
mptHi cm-t, I thought you guys had disappeared09:11
twbIn the past I've used all three about equally, switching each time one of them pissed me off enough09:11
cjwatsonMostly its resolution is fine now09:11
NCommandercjwatson: the lack of hardcoding causes some issues, I've found that d-i breaks interesting ways if you were to have all the packages listed in reverse order09:11
cjwatsonNCommander: So those should be fixed.  We're *not* going back to hardcoding.09:11
NCommandercjwatson: no disagreement, it was something I found out last week, and haven't investigated far enough to actually file a bug on yet09:12
mptcjwatson, I had talked about this "Reinstall from sync" feature with ev, but I should have briefed you on it too. Sorry about that.09:12
cjwatsonAnyway that's not an issue for people who aren't regenerating their own Packages files entirely from scratch, probably also using something other than apt-ftparchive.09:13
cjwatsonmpt: No worries, I briefed myself minimally from the spec.  What priority does this have?09:13
twbWell debootstrap exited without exploding, remains to be seen if the rest of my script assumes cdebootstrap behaviour anywhere09:14
mptcjwatson, Low priority, but it's something cm-t and Deprofundis decided to work on for their project after talking with didrocks.09:14
NCommandercjwatson: we found the initial bug behavior that debootstrap sometimes falls over if the packaging order isn't exact on a regenerated apt-ftparchive's archive. Fiddling with the order caused more explosions.09:15
twbI'm all excited, because last time I did this LZMA2 support was not available, but now I can use it in my live images, yay09:46
xnoxmpt: I have updates for you. See email. Plus shinny mockups https://wiki.ubuntu.com/Ubiquity/AdvancedPartitioningSchemes#Ubiquity_Mockups09:53
mptcjwatson, the "Connect to the Internet" step could be moved and adjusted without implementing the rest.09:53
twbHmm, http://paste.debian.net/171894/ after http://paste.debian.net/171895/09:58
twbOh, it finally showed up.http://paste.debian.net/171896/09:58
twbNow I can go home, hopeful that tomorrow I can just tell it to install a couple more packages without issue, then just ship it and start drinking09:59
xnoxtwb: that paste deserves framing on the wall =)10:00
* xnox bash: !? twb SMASH: event not found10:01
twbThat's why when using rescue images I usually just leave live-config out entirely and drop straight into single10:02
cm-thi mpt (sorry was talking with our teacher, he checks how projects are)10:10
cm-tyes we was "afk" a bit due to many projects to end in the same short time (…)10:11
cm-tso we are in the month mainly dedicated on this project10:12
twbAnd unlike live-build, that script (inc. LZMA2 mksquashfs) runs in *eight minutes* and reduces a 450MBish chroot down to 75MBish .sq10:12
twbyay minimalism10:12
cm-twe forgot to introduce ourself, but thanx to cjwatson for his help on packaging (and taking care to repeat basic IRC CoC, my teamate are not especially familiar with it).10:17
twbcm-t: btw if you're doing homework, it is best practice to say so up front -- because it determines how IRC denizens help you10:19
=== ara is now known as Guest64238
bdmurraycjwatson: did you want me to change my grub2 merge proposal or were you planning on just doing it?16:51
cjwatsonbdmurray: would you mind?16:55
bdmurraycjwatson: no, not at all16:56

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