Started logging meeting in #debian-edu, times are UTC.
[20:01:09] * pere = Petter Reinholdtsen
[20:01:16] <h01ger> #topic 0. add points to the agenda at http://wiki.debian.org/DebianEdu/Meeting - say hi - and who writes the summary?
[20:01:27] * h01ger = Holger Levsen, volunteer to moderate
[20:01:30] <h01ger> +ing
[20:01:48] <h01ger> and i have to finish at 21 UTC, as i have another meeting then
[20:03:24] <h01ger> if there are no other points i suggest to modify the agenda to first quickly submit all wiki issues as bugs and then priorize / validate their openess them.
[20:04:21] * h01ger fixes his printer cable while waiting for people
[20:04:54] <h01ger> after that i'll file bugs (from http://wiki.debian.org/DebianEdu/Status/Lenny) :)
[20:05:22] * pere is sorry, but he do not have the time left to write the summary
[20:07:12] * vagrantc = Vagrant Cascadian
[20:08:00] <pere> I might be a bit absentminded during the meeting. it is because of my son disturbing me. he is not in bed yet.
[20:08:29] <h01ger> :)
[20:09:13] <h01ger> so shall we go ahead with that agenda? vagrant, can you write the summary? else, i could do that, should take 5min with that agenda :)
[20:09:44] <vagrantc> i can write it
[20:09:52] <h01ger> cool!
[20:09:56] <pere> great.
[20:10:04] <h01ger> #action vagrant will write the summary
[20:10:38] * alfton = alf tonny
[20:10:55] <h01ger> hi alfton
[20:12:30] <alfton> hi h01ger
[20:13:15] <h01ger> sep abach alfton k4x bahner taziden jurg_ lazyb0y snorre finnarne Blankoworld Werner meetbot xoswald law AndrewLee jorgenhg winnie white Tolimar sturle C14r pere_away xerakko dinoex jever yannick d7rt vagrantc zobel kyrvin Stianoh bjarne OleA_ pere danielsan klausade schasi: ping, meeting in progress
[20:13:28] <h01ger> so the agenda is:
[20:13:32] <h01ger> Lenny status
[20:13:32] <h01ger> submit all wiki issues as bugs after validating they are still open
[20:13:32] <h01ger> bugs that need fixing before release
[20:13:32] <h01ger> features needed before release
[20:14:03] <h01ger> #topic 1. file the issues from http://wiki.debian.org/DebianEdu/Status/Lenny into http://bugs.skolelinux.org/
[20:14:54] <kyrvin> Knut Yrvin = kyrvin
[20:15:01] <vagrantc> i don't see it reported elsewhere, but i haven't been able to get a successful install from the cd-lenny-test-i386.iso ... always hangs on updating etcinsvk
[20:15:41] <vagrantc> if someone knows a simple way to disable that, i could actually test some issues
[20:15:46] <pere> I believe some of the issues now listed as 'should' issues are really 'must' issues when we consider the target audience for debian edu, teachers with little time and knowledge about linux.
[20:15:46] <h01ger> hi kyrvin! :)
[20:16:00] <pere> vagrantc: it does not hang, it is just horribly slow.
[20:16:01] <h01ger> pere, please rewrite those
[20:16:30] <vagrantc> pere: slow enough that it appears to be hung
[20:16:31] <h01ger> vagrantc, i suggest you file a bug. sounds like a blocker to me, if its so slow that you perceive it as hanging and broken
[20:16:46] <vagrantc> the last time it took 20 minutes ...
[20:17:54] <h01ger> is this still open: "the partition sizes generated by the installer are too small, which is worked around by extending partitions during installation, but this lead to fsck on first boot because of a misfeature in online ext3 resizing. a fix is needed to make sure the minimum partition sizes are large enough to avoid resizing during normal installs. "
[20:17:55] <pere> vagrantc: yeah. it is svk that is slow.
[20:17:57] <h01ger> ?
[20:18:00] <kyrvin> hi h01ger :) long time
[20:18:20] <pere> h01ger: well, unless you or someone else fixed the minimum sizes for all profiles, I guess so.
[20:18:41] <alfton> hi kyrvin nice to see you in here
[20:19:22] * vagrantc registers for a(nother) bugzilla account
[20:19:33] <pere> vagrantc: svk add / commit of a lot of files (like almost every file in /etc/) is horribly slow.
[20:19:43] <vagrantc> h01ger: the partition sizing issue is definitely still present
[20:20:42] <vagrantc> pere: i wonder if it could be done incrementally? i.e. for dir in /etc/* ; do svk add / foo ...
[20:20:50] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1326 just filed, will now file one about the part sizes
[20:21:04] <h01ger> pere, did you replace those shoulds with musts?
[20:21:22] <kyrvin> nice to be back a little alfton :)
[20:21:27] <pere> h01ger: nope. wanted to make sure we have a common understanding of what issues need to be fixed before doing that.
[20:22:33] <pere> h01ger: for example the iceweasel ssl issue is a must fix for me, because teachers will be unable to find the lwat, cups, and all the other links in http://www/ if they have to click through 7 layers of error messages.
[20:22:52] <h01ger> increase patrition size is #1327
[20:23:29] <pere> h01ger: I believe we are doing it wrong if we need to depend on documentation for people to be able to get their work done.
[20:23:36] <h01ger> pere, "Can be "fixed" easily by documenting it" doesnt convince you?
[20:23:42] <h01ger> i tend to agree
[20:23:45] <pere> nope.
[20:23:53] * klausade = klaus ade johnstad
[20:23:57] <h01ger> hi klausade
[20:24:06] <klausade> totally forgott ...
[20:24:57] <vagrantc> a workaround for the partition resizing bug would be to run fsck on each partition before rebooting ...
[20:25:07] <h01ger> pere, i agree its fairly major, but i would release r0 with it. r1 should have it fixed imo. that firefox/iceweasels ssl handling is a pita, is fairly known :)
[20:25:12] <vagrantc> though getting better sizes in the first place would definitely be a good thing
[20:25:34] <h01ger> vagrantc, i think resizing (as in, increasing the default sizes) is pretty easy
[20:25:38] <h01ger> yup
[20:25:41] <pere> h01ger: it is not known to teachers in norway, at least. :)
[20:26:11] <pere> it is very easy. there are three numbers in the partition rules. min, max and "priority".
[20:26:35] <pere> priority is used to calculate how much of the excess space above the minimum should be allocated to a given partition.
[20:26:59] <vagrantc> the hard part is doing enough test installs of all the various permutations
[20:27:08] <pere> so all one need to do is to install a given combination of profiles, run df -h, and calculate the new minimum sizes.
[20:27:13] <pere> vagrantc: yeah.
[20:27:35] <pere> I suspect there is a bug in partman, making it some times create partitions smaller than the minimum size, but did not verify it.
[20:27:53] <vagrantc> if i could figure out a simple workaround for the svk issue, i could do a few test installs to get a better idea.
[20:28:00] <pere> it would explain some installs that confused me.
[20:28:36] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1327
[20:28:37] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1328
[20:28:38] <h01ger> filed
[20:28:55] <pere> vagrantc: rm /usr/bin/svk; ln -s /bin/true /usr/bin/svk should work.
[20:29:24] <vagrantc> i guess i could log in and do that manually...
[20:29:45] <pere> vagrantc: it should be done during installation, before tasksel is executed.
[20:29:49] <vagrantc> right
[20:29:56] <h01ger> vagrantc, can you file a bug about this please, unless you commit a fix for this of course ;)
[20:30:09] <vagrantc> h01ger: i'm still waiting for my bugzilla password
[20:30:18] <h01ger> vagrantc, when did you apply?
[20:30:31] <pere> what should we do if svk is still slow when we want to release? I have not seen any progress with svk in debian.
[20:31:11] <vagrantc> h01ger: sometime between the start of the meeting and now...
[20:31:18] <pere> the etcinsvk feature is very valuable, but it slow down the installation _a lot_.
[20:31:32] <vagrantc> what about gitkeeper ?
[20:31:37] <vagrantc> or etckeeper ?
[20:31:38] <pere> vagrantc: please report it to debian too.
[20:31:46] <h01ger> vagrantc, :)
[20:31:55] <pere> etckeeper is interactive, and do not seem to be a good alternative.
[20:32:03] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1329
[20:32:16] <pere> etcinsvk is only recording changes, and not prompting the admin for comments every time packages are upgraded.
[20:32:17] <h01ger> what is the status: "pkgsel failes to install octave cousing a error in the installer, when installing a thin client server profile(perhaps also workstation?) . rerunning the default selected step works. (pere: FIXED?)"
[20:32:20] <h01ger> ?
[20:32:49] <pere> h01ger: did not test recently. anyone else?
[20:33:54] <pere> the profile question look really crappy in the text installer, btw. there is a lot of blank lines, and the options are put in a too small box forcing users to scrol to find the standalone option.
[20:34:02] * h01ger assumes "LDAP server do not accept encrypted connections (#473796) due to gnutls integration issue.
[20:34:02] <h01ger> Does at least work with PHP/LWAT and ldapsearch - so what is the issue left? " is closed. that bug is closed in debian lenny, so. it's closed. and removing from that list is pretty safe, as it will be noted again, if there is an issue :)
[20:34:08] <pere> the expert profile question look worse.
[20:34:14] <h01ger> pere, are our images based on latest d-i?
[20:34:22] <pere> h01ger: not sure. expect so.
[20:34:28] <h01ger> there was a related fix recently
[20:34:40] <h01ger> recently = within the last 5 days
[20:34:47] * pere did not test php/lwat.
[20:35:03] <pere> I did test swedish standalone install today, and there the self test claimed cups wasn't running.
[20:37:19] <h01ger> was kde set to swedish too?
[20:38:15] <pere> yes. added a comment about this in bugzilla.
[20:38:26] <pere> found time to log in just before I left work.
[20:39:31] <h01ger> ah :)
[20:39:37] <pere> I guess we should have a list of priority langauges to check. norwegian bokmål and nynorsk, german, french, spanish and what more?
[20:39:45] <h01ger> then taht bug can be closed.
[20:39:53] <h01ger> italian
[20:39:59] <pere> perhaps a test plan / description on what to test on the wiki will help people test?
[20:40:00] <h01ger> the manual is fully translated
[20:40:11] <h01ger> pere, i think so. dont we have one already?
[20:40:24] <pere> h01ger: not sure. could be. me is a bit absent minded at the moment.
[20:40:27] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1330
[20:40:43] <h01ger> http://wiki.debian.org/DebianEdu/HowTo/TestCDinstall
[20:40:46] <h01ger> there :)
[20:41:18] <h01ger> #idea a test plan / description on what to test should be added to http://wiki.debian.org/DebianEdu/HowTo/TestCDinstall
[20:41:56] <pere> h01ger: It explains how to test. I thought more on 'installl standalone in different langauges, and verify that kdm, kde is translated and the correct spell checker is enabled in kde and openoffice.org.
[20:42:43] <h01ger> pere, yes. that sounds exactly what can be added at the bottom / middle of that page?
[20:43:14] <pere> and lwat, sitesummary, cups, ltsp etc should be tested too.
[20:43:16] <h01ger> http://wiki.debian.org/DebianEdu/HowTo/TestCDinstall is linked http://wiki.debian.org/DebianEdu/Status/Lenny on top :)
[20:43:42] <h01ger> #idea the test plan should include lwat, sitesummary, cups, ltsp etc too
[20:44:12] <pere> please do not wait for me. I just do not manage to do all I would like to do these days. Not that I ever did, but I manage even less these days.
[20:45:00] * h01ger nods
[20:46:21] <h01ger> thats very fine, i understand that having a baby is a wonderful timesink. i just wonder how to deal (in general, nothing specific atm) with stuff where you say its a "must" feature/fix but you cant do it. other volunteers cant be forced neither...
[20:46:42] <h01ger> but anyway, "usplash do not work with single user mode (#431560)" seems solved to me, that bug is closed in lenny...
[20:47:06] <pere> h01ger: btw, regarding the iceweasel ssl issue. do you not believe we will loose all new users of r0 if they are forced to jump through all the error messages of iceweasel to be able to add a user in ldap?
[20:47:13] <h01ger> "usplash and kdm to not work reliably together. Some times, kdm fail to start during boot due to some VT switching issue (#489734). (pere: FIXED?)" - the same.
[20:47:16] * h01ger removes both
[20:47:34] <pere> good.
[20:47:46] <h01ger> pere, not if we label the cd with "use konqueror for system administration"
[20:48:03] <h01ger> pere, thanks to you for fixing sysvinit :)
[20:48:19] <h01ger> "Samba do not start, according to the testsuite test." - still open?
[20:48:47] <pere> btw, regarding usplash, I suspect the new usplash require huge amount of memory, and thus making thin clients require huge amount of memory. we might want to drop usplash from the thin client boot.
[20:49:05] <pere> h01ger: not even if we label the Cd like that. :)
[20:49:11] <pere> h01ger: no idea.
[20:49:31] <pere> the testsuite report heaps and heaps of errors these days, at least a week or so ago.
[20:50:55] * h01ger nods
[20:51:19] <pere> most of it because of missing task packages.
[20:51:22] <vagrantc> new cd images were generated just 2 days ago
[20:52:11] <h01ger> vagrantc, they should be done daily?
[20:52:12] <pere> vagrantc: should be generated at least every mirror pulse (every day). something must be wrong.
[20:52:22] <h01ger> http://bugs.skolelinux.org/show_bug.cgi?id=1331
[20:53:48] <h01ger> "OCS Inventory server is not installed properly, and the client do not automatically know how to connect to the server (#489461)."
[20:53:48] <vagrantc> h01ger, pere: hmmm... i definitely recall not seeing new images since november 18th or so until about 2 days ago
[20:53:57] <h01ger> #489461 is fixed, actually
[20:54:31] <h01ger> vagrantc, can you login to a.s.o and investigate or dont you have an account?
[20:54:44] <vagrantc> ftp://ftp.skolelinux.org/cd-lenny-test-dvd/ is dated 12/01/2008
[20:54:44] <pere> h01ger: I doubt it. the "fix" was to fix the package installation failure, not to fix the installatin.
[20:54:51] <h01ger> #topic 2. other business
[20:54:56] <vagrantc> h01ger: i'm not sure if i have an account
[20:54:58] <h01ger> meeting time is almost over
[20:55:22] <h01ger> #idea we need to finish (filing and) prioritising bugs at the next meeting
[20:55:38] <h01ger> vagrantc, /me checks
[20:55:43] <h01ger> any other business?
[20:55:49] <pere> h01ger: we want the ocs invetory server installed and _configured_, which require a running mysql at installation time within d-i. it is obviously not happening yet.
[20:55:55] <h01ger> there will be a developer meeting in 2 days?!
[20:56:08] <pere> h01ger: yes, but very few have announced their participation.
[20:56:26] <pere> me, alfton and akai are the only ones listed on the wiki.
[20:56:38] <h01ger> pere did you look at the fix for #489461 ? i'm not sure your statement (about ocs) is still true
[20:56:44] <h01ger> pere, ui. pity
[20:56:51] <kyrvin> I believe that a bit more will come since it's in Oslo. In the same breath, we are in December.
[20:56:54] <pere> h01ger: not recently.
[20:56:56] <h01ger> some sugar people are pondering coming in january
[20:57:18] <kyrvin> Then people are more occupied with X-mas planning and shoping
[20:57:22] <pere> kyrvin: well, they might meet closed doors unless coordinating with alfton and akai.
[20:58:03] <kyrvin> What we usually do is to put up a sign, as has been there at every gathering when I'm comming 11 instead of 10
[20:58:05] <pere> h01ger: yes. "allow to install" just mean the postinst script do not fail. the package is not configured.
[20:58:11] <h01ger> #topic 3. next meeting
[20:58:22] <h01ger> pere, then we can use the firstboot script, cant we?
[20:58:29] <pere> kyrvin: well, it is not working that way on friday, at least.
[20:58:32] <h01ger> in two weeks, on the 17th?
[20:59:05] <pere> h01ger: sure, or a script starting mysql within d-i, configuring and then shutting down mysql. just need someone to _do the work_ and _get it to function_.
[20:59:19] <kyrvin> I though it was closed on friday, and we would start on saturday because Werner is delivering a key to one in Trondheim?
[20:59:25] <h01ger> pere, why with d-i, why not on firstboot?
[20:59:38] <h01ger> next meeting in two weeks, on the 17th?
[20:59:39] <pere> h01ger: to make sure the system is ready to go when it is turned on?
[20:59:45] <kyrvin> kyrvin: of topick and should wait
[20:59:50] <h01ger> pere, we have a firstboot script already
[21:00:11] <pere> h01ger: sure. I leave those decisions to those working on fixing it.
[21:00:32] <h01ger> #action me decides to hold another meeting in two weeks. we can still change that anyway
[21:00:41] <pere> I would prefer the firstboot script to only do testing, and only for test builds.
[21:00:45] <h01ger> ah
[21:00:50] <h01ger> i was about to ask :)
[21:01:05] <pere> it would allow the readahead profiling run to run on the first boot, and not on the second.
[21:01:11] * h01ger thanks everybody for attending and hopes even more people will attend in 14 days :)
[21:01:15] <h01ger> #endmeeting
Information on meetbot is available at meetbot.debian.net