20080416

Agenda

Items we will be discussing:

  • Review ACTION points from previous meeting.
  • should the server upgrader run with "dpkg --force-overwrite" to ignore file conflicts caused by broken Replaces in the packages? Currently it does run without it. The desktop upgrader runs with --force-overwrite. The implication is that the upgrade may stop because of a missing/bad replaces (quite common unfortunately) and require manual cleanup.
  • Review high priority bugs related to the Server Team.
  • Ubuntu Documentation Portal (DustinKirkland)

  • White paper proposal (nijaba)
    • Start a community effort to turn select how-to from the wiki into official white papers to make them more appealing to corporate type.
  • Review ServerTeam/Roadmap and ServerTeam/ReportingPage.

  • Agree on next meeting date and time.

Minutes

server upgrader run with --force-overwrite

mvo explained that update-manager is running with --force-overwrite when doing a desktop upgrade. He wanted to know it was a good idea that do-release-upgrade also run with --force-overwrite on a server upgrade. He explained that --force-overwrite makes dpkg ignore errors about missing Replaces: when files from one package travel into another one for example which is unfortunately pretty common. After some discussion, the server team agreed with mvo that running the server upgrader with --force-overwrite makes sense.

mathiaz reminded that important bugs should be milestoned for hardy. If anyone is not able to milestone a bug, he should get with a developer who will have a quick look at the bug and milestone it if appropriate.

nijaba wondered whether kvm testing has been tested enough. mathiaz reported a couple of issues he ran into during his daily usage. Unfortunately these were not easily reproducible or he found workarounds for them.

nealmcb mentioned that soren, our virtualization specialist, was already virtually present on all machines. soren didn't deny it and we all started to chase down soren virtual presence in our machines. All searches were unsuccessful though.

kirkland put up a page [1] using Google Custom Search that searches different resources related to Ubuntu. It includes about a dozen ubuntu/launchpad/etc related websites which are given priority when doing a global Google search of the web. While designing the page he was unable to find an online database of man pages shipped in ubuntu releases. He talked with cjwatson to put together a site of manpages. mathiaz suggested to contact the Documentation Team and present them his work.

[1]: http://people.ubuntu.com/~kirkland/search.html

White paper proposal

nijaba presented an idea to turn wiki how-tos into white papers so that they are more appealing to corporate types. He suggested to start with the content of help.ubuntu.com and see which pages could be turned into white papers. mathiaz suggested to contact the Documentation team and start by defining what would be the content of a whitepaper. Having a session at UDS to cover that would also be a valid option.

Agree on next meeting date and time

Next meeting will be on Wednesday, April 23th at 21:00 UTC in #ubuntu-meeting.

Log

[23:01:42] <mathiaz> Today's agenda: https://wiki.ubuntu.com/ServerTeam/Meeting
[23:01:52] <owh> It's always good to know we're loved by the bot :)
[23:02:06] <mathiaz> [TOPIC] Review ACTION points from previous meeting
[23:02:17] <mathiaz> https://wiki.ubuntu.com/MeetingLogs/Server/20080409
[23:02:33] <mathiaz> There was one action - and I completed it
[23:02:45] <ajmitch> ah, another meeting
[23:02:54] <zul> hey ajmitch
[23:03:05] <mathiaz> the samba-server task will install libpam-smbpass
[23:03:29] <mathiaz> which will keep the user password synced between samba and the system
[23:04:21] <mathiaz> [TOPIC] server upgrader run with --force-overwrite
[23:04:33] <soren> ...how does it install itself into pam?
[23:04:35] <mathiaz> mvo: could you give an overview of your request ?
[23:04:44] <jdstrand> soren: pam itself is updated
[23:05:01] <mathiaz> soren: there is an optional pam_smpass in pam
[23:05:11] <jdstrand> soren: password optional pam_smbpass.so nullok use_authtok use_first_pass
[23:05:17] <soren> Oh, really? That was easy :)
[23:05:24] <jdstrand> soren: that is by default
[23:05:32] <mvo> mathiaz: sure
[23:05:36] <soren> jdstrand: Clever.
[23:05:47] <mvo> this is about if we should run the upgrader with dpkg --force-overwrite by default or not
[23:06:00] <mvo> the desktop runs with --force-overwrite, the server does not (currently)
[23:06:03] <jdstrand> soren: well, and pam_unix.so is now requisite
[23:06:40] <mvo> --force-overwrite will make dpkg ignore errors about missing Replaces: when files from one package travel into another one for exmaple. its unfortunately pretty common
[23:06:54] <mvo> currently a overwrite problem makes the upgrade abort and the admin needs to clean it up manually
[23:07:26] <jdstrand> mvo: how common is 'pretty common'?
[23:07:29] <soren> Seems reasonable enough to me.
[23:07:30] <mvo> I'm not sure if that is woth it, most admins will clean up by running --force-overwrite by hand and continue the upgrade, there is not too much that can be done else
[23:07:47] <nealmcb> what is an example of a package conflict like this?
[23:07:57] <mvo> jdstrand: our current ubuntu-desktop -> ubuntu-desktop upgrade from dapper->hardy has two file conflicts
[23:08:10] <mvo> we are fixing them, but I'm sure there are plenty in universe left
[23:08:16] * jdstrand nods
[23:08:36] <soren> mvo: What are the possible dangers?
[23:08:36] <jdstrand> mvo: it seems reasonable to have consistent behavior between desktop and server
[23:09:00] <soren> mvo: It's only really a problem if a file didn't actually move, but is in both packages, correct?
[23:09:02] <mathiaz> mvo: right - so the question is whether server users should see an upgrade fail when they installed packages that are problematic
[23:09:08] <mvo> soren: when we enable it? I don't see any real dangers, the file overwrite thing is recorded in the logs
[23:09:30] <mvo> soren: yes, that is a real error then, I don't think we have a lot of those, most I have seen are transitional ones
[23:09:31] <nealmcb> and people may need to get it from backup?
[23:09:45] <mvo> mathiaz: exactly
[23:09:46] <mathiaz> mvo: the reason to enable it on desktop is to avoid leaving end users in a state where the upgrade failed ?
[23:09:49] <soren> mvo: Yeah, that's my feeling, too. IMO --force-overwrite is fine.
[23:10:21] <mvo> mathiaz: yes, the reason is that on the desktop it may stop in the middle of a X upgrade and when the user panics and reboots his machine will not come up with X anymore
[23:10:43] <mvo> and because it is silly to stop a upgrade because of this IMHO
[23:11:08] <owh> Can we ask the user during upgrade?
[23:11:19] <nealmcb> how does the script determine that it is a "server"?
[23:11:21] <jdstrand> well, dpkg is careful so as not to destroy one's system
[23:11:23] <owh> I mean it's not like a server is running X?
[23:11:23] <mvo> seriously, there is not too much that can be done to fix it other than to either remove the package (problematic in a lot of cases) or to fix it manually with --force-overwrite
[23:11:27] <mathiaz> mvo: I agree - I just want to make sure that --force-overwrite will only deal with missing Replaces
[23:11:49] <mvo> nealmcb: we have a special mode flag in the upgrader, the server upgrade behaves traditionally slightly different
[23:11:51] <nealmcb> and how do we expect the packages to be fixed?
[23:11:51] <jdstrand> mvo: is it possible to have --force-overwrite by default, but have the option to disable it?
[23:12:02] <mathiaz> mvo: At first, force-overwrite seemed more generic - it may overwrite anything...
[23:12:02] <mvo> owh: unfortunately we have not the means for asking currently
[23:12:06] <nealmcb> mvo: based on kernel version or something?
[23:12:38] <nealmcb> or command-line vs gui usage?
[23:13:00] <mvo> jdstrand: it can be added (and is probably a good idea). I have no switch readily available though (need to add code)
[23:13:08] * jdstrand nods
[23:13:16] <zul> mvo: what if the users have x installed with apache and the likes?
[23:13:20] <mvo> nealmcb: do-release-upgrade will default to server, update-manager will default to desktop, both have a --mode switch to overwrite
[23:13:30] <nealmcb> ahh - got it
[23:13:47] <mvo> zul: if he runs update-manager he gets the desktop, if he runs do-release-upgrade he gets the server one
[23:13:55] <zul> mvo: ok
[23:14:07] <jdstrand> mvo: IMO I agree --force-overwrite is a good default, but on a server it seems reasonable that someone may have customized packages/configurations/etc where they would not want --force-overwrite by default
[23:14:10] <owh> mvo: That's not very intuitive I'd have to say.
[23:14:46] <mathiaz> jdstrand: well - configuration files should not be overwritten with --force-overwritte
[23:14:48] <owh> mvo: Does the user/admin get fair warning that they are doing a server or desktop upgrade?
[23:14:51] <mvo> owh: I'm happy about suggestens to improve that for intrepid (too late for hardy, sorry)
[23:15:12] <nealmcb> Can packages be automatically audited for these conflicts, or are they sometimes set dynamically?
[23:15:13] <jdstrand> mathiaz: on no, force-overwrite doesn't deal with that AFAIK
[23:15:52] <mvo> jdstrand: right, I will look into what we can do about a switch for this
[23:16:10] <owh> Hold on, we're not talking about overwriting config files with --force-overwrite -- so the files that we're overwriting are poorly created packages - am I understanding this correctly?
[23:16:11] <jdstrand> I was more thinking of a 3rd party or inhouse deb
[23:16:15] <mathiaz> jdstrand: ah - so you think about someone that has a custom-build apache2 and then suddendly on an upgrade files are overwritten ?
[23:16:19] <mvo> nealmcb: there is a conflicts-checker script in development
[23:16:24] <nealmcb> :-)
[23:16:25] <jdstrand> mathiaz: exactly
[23:16:43] <soren> mvo: I'm not sure what --force-overwrite should force other than missing Replaces: ?
[23:16:54] <mathiaz> jdstrand: ok - I'd say that if you build your own apache2 deb , then you'd probably test the upgrade
[23:17:06] <mathiaz> jdstrand: and not blindly dist-upgrade
[23:17:15] <jdstrand> ideally, yes
[23:17:24] <soren> mvo: You seem to be suggesting there's another situation that that switch makes dpkg override (or force or whatever you want to call it)?
[23:17:30] <jdstrand> I am just erring on the side of caution-- I agree it is a good default
[23:17:38] <mvo> soren: you are the dpkg maintainer ;) I don't know of anything else too
[23:17:49] <soren> mvo: Ok, good. I was trying to think of one, but couldn't :)
[23:17:52] <zul> jdstrand: if you compile your own stuff then you should know to rebuild your own stuff
[23:18:04] <mvo> soren: sorry if I gave this impression
[23:18:06] <owh> zul: Exactly.
[23:18:07] <jdstrand> that was just one example
[23:18:09] <soren> mvo: No worries :)
[23:18:29] <owh> zul: Some warning that your package got butchered would be good though.
[23:18:35] <nealmcb> it all seems scary but I defer to the experts....
[23:18:43] <owh> zul: s/package/files/
[23:18:44] <soren> mvo: If there indeed was another I would have just made a new switch to dpkg to only force this situation.
[23:18:47] <zul> owh: probably too late for hardy :)
[23:18:57] <soren> mvo: That's why I was asking.
[23:18:59] <mathiaz> mvo: ok - so if force-override is just about Replaces, it makes sense to enable it on server too
[23:19:15] <owh> mvo: So, if you do the --force-override, is there a log to show you what happened?
[23:20:04] <mvo> owh: yes, theere is /var/log/dist-ugprade/apt-term.log that shoudl show exactly what happend
[23:20:14] <mvo> including the warnings from dpkg aobut file overwrite problems
[23:20:25] <mathiaz> mvo: so it's a +1 for me
[23:20:30] <owh> mvo: Well, then we can support them when they run into problems. +1
[23:20:35] <nealmcb> this is force-overwrite, right? not force-override?
[23:20:48] <nijaba> I think +1 if we document this in the release notes
[23:20:59] <owh> I agree with nijaba
[23:21:03] <mvo> *cough* yes, --force-overwrite
[23:21:07] * faulkes- does as well
[23:21:09] <soren> mvo: I think this is kind of like apport in some sense. We keep it enabled for most of the dev cycle to iron out the creases and disable it for the final release to reduce noise we can't do much about anyway.
[23:21:21] * mvo nods
[23:21:25] <mvo> ok, thanks a lot for the feedback!
[23:21:35] <soren> THanks for upgrading our servers!
[23:21:38] <soren> :)
[23:21:40] <mathiaz> you're welcome mvo :)
[23:21:42] <owh> mvo: Thanks for the work :)
[23:21:51] * nijaba hugs mvo
[23:22:03] <mathiaz> Let's move on
[23:22:10] <mathiaz> [TOPIC] Review high priority bugs related to the Server Team
[23:22:15] * nealmcb hopes the editors can overwrite the override with overwrites ....
[23:22:34] <mathiaz> so - rc is around the corner
[23:22:37] * faulkes- underwrites nealmcb
[23:22:40] <mathiaz> ie - tomorrow :D
[23:22:49] <nijaba> mathiaz: no, delayed to friday
[23:22:56] <zul> mathiaz: friday now
[23:23:03] <mathiaz> and final will be there next week
[23:23:13] <mathiaz> nijaba: ok
[23:23:41] <mathiaz> dendrobates wants to know if anyone came across bugs that should be targeted for 8.04
[23:23:59] <mathiaz> make sure that they're milestoned in LP
[23:24:18] * soren goes to file bugs..
[23:24:32] <owh> ROTFL
[23:24:51] <nijaba> there are a few people reporting issues with kvm under heavy load. Did we test this enough?
[23:25:16] <kirkland> nijaba: issues with vm clients, or the server?
[23:25:16] <mathiaz> nijaba: well - I'm using kvm to do my iso testing
[23:25:26] <mathiaz> I ran into a couple of issue
[23:25:28] <soren> nijaba: Hard to say, really.
[23:25:35] <nijaba> server getting a lot of traffic
[23:25:49] <mathiaz> soren: sometimes when the guest reboots kvm dies
[23:25:50] <soren> nijaba: I have a few instances running "stress --hdd 3 --io 3 --cpu 3" and they're all happy.
[23:26:13] <nijaba> soren: what about testing with apache for example?
[23:26:23] <soren> nijaba: I'm not so much as it's slowing my machine down, but hey... that's what you get for torturing your dev box like that.
[23:26:36] <jdstrand> mathiaz: really? I have never seen that
[23:27:06] <soren> mathiaz: Sounds rather suboptimal. I trust you have a bug number for me? :)
[23:27:08] <mathiaz> jdstrand: yop - happened a couple of times today
[23:27:20] <mathiaz> soren: I was about to ask where I should look for more information
[23:27:36] <owh> mathiaz: LP :_)
[23:27:37] <nijaba> mathiaz: the test you did with dovecot, they were under kvm?
[23:27:53] <mathiaz> and right now three of my guests are sutck on 'Booting from hard drive'
[23:28:04] <mathiaz> after the install
[23:28:18] <mathiaz> nijaba: at first they were under kvm
[23:28:31] <mathiaz> nijaba: then I moved to real hardware
[23:28:47] <nijaba> mathiaz: so I guess this qualifies to kvm stress testing...
[23:29:16] <mathiaz> I've also come accross some issue with networking - when I add six guests running at the same time, bringing a 7th up - its networking would not work
[23:29:39] <mathiaz> nijaba: right - that's one sort of stress testing
[23:29:40] <kirkland> mathiaz: what are their netmasks?
[23:29:54] <mathiaz> kirkland: /24
[23:30:02] <jdstrand> mathiaz: we talked about the networking-- it may be bridge related
[23:30:11] <jdstrand> I've not seen it
[23:30:14] <kirkland> mathiaz: hrm, that's not it then ;-)
[23:30:31] <mathiaz> I ran in these issues over the last weeks - but was chasing down other bugs - so I haven't taken the time to debug the issue
[23:30:45] <jdstrand> mathiaz, soren: that said, I believe kees uses a bridged network, and he had 10 vms running concurrently yesterday
[23:30:55] <mathiaz> soren: I've also seen libvirtd dying when I tried to load 5-10 guests at the same time IIRC
[23:31:26] <mathiaz> jdstrand: OTOH the networking issue may be related to a dhcp configuration I've fixed today on my network
[23:32:38] <owh> mathiaz: There would be some logs for the DHCP server to show that perhaps?
[23:32:39] <mathiaz> so - make sure that bugs that are important are milestoned for hardy
[23:33:18] <mathiaz> and if don't have permission to milestone them, msg me or dendrobates or another member of ubuntu-dev so that we can take a look at it
[23:33:53] <mathiaz> we'd better spend some time looking at a bug rather than having one splipping unoticed in final
[23:33:57] <soren> mathiaz: Dude, you need to tell me these things! :)
[23:34:19] <soren> jdstrand: Yeah, I've had ~10 running at the same time too. :/
[23:34:47] <mathiaz> soren: yes - I just hadn't taken the time to start debugging it
[23:35:02] <nealmcb> soren: we thought the virtualization specialist was already virtually present on all machines....
[23:35:44] <soren> nealmcb: I may very well be..
[23:35:48] <nealmcb> lol
[23:35:54] <soren> But admitting to it would blow my cover.
[23:35:56] <faulkes-> sudo apt-get install soren
[23:36:02] <owh> For those interested, I just jumped through the roof with a thunderclap right outside my window.
[23:36:04] <soren> boom!
[23:36:07] <nijaba> soren is watching you
[23:36:14] * jdstrand peeks inside his vms looking for soren
[23:36:39] <owh> jdstrand: You know the little man with the hammer? - it's soren :)
[23:36:40] <mathiaz> so let's stop chasing soren in ours vms and move on
[23:36:40] <nealmcb> deep, below the microcode....
[23:36:52] * jdstrand actually likes thinking soren is inside my vms-- just a click away!
[23:36:59] <mathiaz> [TOPIC] Ubuntu Documentation Portal
[23:37:02] <soren> This is getting out of hand..
[23:37:14] <mathiaz> kirkland: what's your brilliant idea ?
[23:37:14] <soren> :)
[23:37:18] <faulkes-> the stalker becomes the stalkee
[23:37:20] <kirkland> mathiaz: http://people.ubuntu.com/~kirkland/
[23:37:34] <kirkland> mathiaz: it's using Google's Custom Search Engine technology
[23:37:43] <nijaba> This is great stuff !
[23:37:59] <kirkland> mathiaz: I've listed about a dozen ubuntu/launchpad/etc related websites which are given priority when doing a global Google search of the web
[23:38:08] <kirkland> mathiaz: furthermore, once you have hits, you can refine your results
[23:38:15] <owh> kirkland: I like when you search for 'samba', you get some google ads on the side :)
[23:38:28] <kirkland> owh: for $100/year, you can pay google to remove the ads
[23:38:47] <kirkland> owh: i'd think Canonical might do so, if we decide this is generally useful
[23:38:48] <owh> kirkland: Me personally, or the search provider?
[23:39:00] <kirkland> i thought about coughing it up myself
[23:39:11] <kirkland> but figured i'd see if Canonical might first ;-)
[23:39:18] <kirkland> mathiaz: now the cool part....
[23:39:23] <kirkland> mathiaz: is in the "Refinements"
[23:39:41] <faulkes-> I should note, I'm about half-way through a rewrite of the forum code so it provides dropdown -> jump to ability for the forum-stats stuff
[23:39:43] <kirkland> mathiaz: once you have hits, you can click the little links over the hits and only search certain documentation providers
[23:39:49] <owh> Ooh, nice.
[23:39:50] * faulkes- has been swamped with work lately though
[23:39:57] <sommer> that is sweet!
[23:40:05] <kirkland> there's a logical order that i've defined there
[23:40:19] <kirkland> which is what I believe to be ordered from most-authoritative/informative
[23:40:35] <kirkland> acutally, "authoritative" isn't the right word....
[23:40:37] <kirkland> Manuals > Official Docs > Wiki Docs > Answers > Forums > Lists > Bugs > Code
[23:40:49] <owh> WFM
[23:40:53] <kirkland> i generally start with manpages/infopages
[23:40:57] <kirkland> then official documentation
[23:41:01] <kirkland> then wikis/community docs
[23:41:06] <kirkland> then Launchpad answers
[23:41:08] <kirkland> then forums
[23:41:11] <kirkland> then mailing lists
[23:41:23] <kirkland> at which point, Launchpad Bugs are probably next
[23:41:31] <kirkland> and finally, down into the Source Code
[23:41:44] <nealmcb> cool - seems like bugs might be higher
[23:41:44] <kirkland> there are two weaknesses, currently, as I see it....
[23:41:49] <owh> kirkland: Can you add packages as such a link?
[23:41:51] <mathiaz> kirkland: that's great stuff !
[23:42:02] <jdstrand> kirkland: cool!
[23:42:06] <owh> +1
[23:42:14] <kirkland> first, there is no authoritative site that indexes all Ubuntu Manpages that I know of
[23:42:18] <kirkland> if anyone knows, pray tell
[23:42:30] <owh> kirkland: No, but there is a debian package that does.
[23:42:31] <kirkland> linuxmanpages.com seems to show SUSE (or maybe RH) manpages
[23:42:37] * owh forgets its name.
[23:42:37] <kirkland> and yes, there is manpages.debian.net
[23:42:48] <nijaba> goobuntu.com
[23:42:50] <kirkland> so I'm working with cjwatson to put together a site of manpages
[23:43:01] <nijaba> but not as nice
[23:43:01] <kirkland> for starters, I built this: http://ubuntu.dustinkirkland.com/manpages/
[23:43:16] <owh> kirkland: No, I mean, a .deb that indexes installed man pages and makes them into a website.
[23:43:16] <kirkland> which contains flat text files of all manpages in Dapper - Hardy
[23:43:28] <kirkland> owh: ah, cool. we need that for Ubuntu
[23:43:37] <kirkland> mathiaz: todo? research that magic deb?
[23:43:37] * owh is just looking for it.
[23:44:06] <kirkland> cjwatson had the idea to use richer text/html markup for the manpages, which would be quite nice
[23:44:19] <kirkland> mathiaz: and then hopefully get it hosted on help.ubuntu.com somewhere
[23:44:26] <owh> kirkland: man2html
[23:44:27] <kirkland> mathiaz: and automatically kept uptodate
[23:44:31] <kirkland> owh: nice
[23:44:39] <kirkland> I'll rerun my script using that
[23:44:40] <nijaba> search.ubuntu.com would be nice
[23:44:48] <owh> Yes
[23:44:56] <kirkland> it takes about 6 hours to do all of the distros
[23:45:04] <cjwatson> man2html is a bit curious
[23:45:05] <kirkland> i'll rerun with man2html tonight, thanks owh
[23:45:08] <mathiaz> kirkland: that seems like a good plan
[23:45:19] <kirkland> cjwatson had some more troffy ideas, i think
[23:45:24] <cjwatson> as the groff maintainer I'm kind of leery of schemes to parse manual pages without using groff
[23:45:28] <owh> kirkland: It's not the actual app that I was looking for. I'll see what I can dig up from my archive.
[23:45:32] <kirkland> pardon me, groffy :-)
[23:45:39] <cjwatson> but, I recognise that it might be a decent short-term solution
[23:45:40] <nijaba> kirkland: so we should send you our improvement requests, right?
[23:45:42] <mathiaz> kirkland: once the man page in html are online, google will pick them up and they should be searchable
[23:46:00] <kirkland> nijaba: yes, please do
[23:46:11] <kirkland> oh, internationalization is the other shortcoming
[23:46:25] <mathiaz> kirkland: right - so that is a great idea IMO
[23:46:33] <mathiaz> kirkland: you should get in touch with the documentation team
[23:46:34] <cjwatson> manual pages are localisable ...?
[23:46:37] <jdstrand> kirkland: this is really excellent idea
[23:46:38] <kirkland> mathiaz: yes, you have to tell google to index your site
[23:46:43] <nijaba> well, if your index.html is available, loco could set it up
[23:46:54] <mathiaz> kirkland: I'm sure you'll find a lot of interested people there.
[23:47:03] <kirkland> mathiaz: hwere?
[23:47:04] <nealmcb> kirkland: and what about all the code - is it in one place?
[23:47:05] <nijaba> kirkland: actually, it will go faster if google finds a link to it
[23:47:17] <kirkland> nealmcb: the code is one small HTML file
[23:47:30] <kirkland> nealmcb: and the rest is in a Google Gadget
[23:47:36] <nealmcb> sorry - I mean all the code to all the packages...
[23:47:37] <kirkland> kirkland: i can add others to the project
[23:47:39] <mathiaz> kirkland: https://wiki.ubuntu.com/DocumentationTeam
[23:47:42] <kirkland> nealmcb: oh
[23:47:59] <nijaba> kirkland: talking to yourself?
[23:48:07] <mathiaz> so let's move on
[23:48:15] <kirkland> mathiaz: okay, thanks.
[23:48:20] * nealmcb cheers for kirkland
[23:48:22] <owh> kirkland: Dunno if you saw my question earlier. Can you add a link to "Packages" there too?
[23:48:26] <mathiaz> kirkland: great work BTW ! :)
[23:48:30] <mathiaz> [TOPIC] White paper proposal
[23:48:32] <owh> That's very nice.
[23:48:33] <cjwatson> (with regard to localised manual pages, one of my specific concerns is that I've done a lot of work over the last year to cope with character encodings in manual pages correctly - it's quite a thorny issue for historical reasons - and I suspect tools that don't involve groff/man will get it wrong for non-English pages)
[23:48:34] <kirkland> owh: sure, msg me private
[23:48:37] <mathiaz> nijaba: ?
[23:49:15] <nijaba> soren, I'd like to submit the idea to turn wiki howto into wp
[23:49:29] <nijaba> so that they are more appealig to corporate types
[23:49:38] <nijaba> s/soren/so
[23:50:16] <nijaba> it would be a nice way to define that it has been validated by us
[23:50:39] <sommer> do you have a list of topics?
[23:51:15] <nijaba> well, my suggestion was to go to help.ubuntu.com and sort out which one would be fitting
[23:51:19] <nealmcb> nijaba: what sort of workflow? i.e. how to sync when further editing is done on either the wiki pages or the whitepapers
[23:52:16] <nijaba> nealmcb: wp are not docs, they are just a way to tell people about great features and how to implement them.
[23:52:38] * sommer needs to read more white papers
[23:52:56] <nijaba> so I would think that what counts is not how up to date they are, but how wide a coverage they provide
[23:53:15] <faulkes-> I would have to agree with nijaba
[23:53:52] <faulkes-> I can talk all day to manglement here, point them at docs but what gets hammered home most, is a good white paper, short, sweet, informative
[23:53:59] <nijaba> but I think it would be a nice way to elevate some of the great how to we have and give them some public light
[23:54:08] * faulkes- nods
[23:54:22] <mathiaz> nijaba: right - the idea is interesting
[23:54:31] <faulkes-> exposure to drive adoption
[23:54:38] <mathiaz> nijaba: I'm not sure if the content you'd like to see would be the same as how-to
[23:54:50] <nealmcb> [reminder - kubuntu in 5 minutes....]
[23:54:54] <mathiaz> nijaba: I'd suggest to contact the documentation team (again) ;)
[23:54:59] <nijaba> mathiaz: it will have to be adapted to alway provide a high level overview
[23:55:02] <kirkland> nijaba: i've some experience with this, i'll ping you later
[23:55:10] <mathiaz> nijaba: and as nealmcb just said, we're running out of time :)
[23:55:14] * faulkes- has to head out
[23:55:40] <mathiaz> nijaba: right - so I think we should define what would be the content of a "whitepaper"
[23:55:51] <sommer> nijaba: I'm also interested in helping, just let me know :)
[23:55:53] <nealmcb> is this a good uds topic?
[23:55:56] <mathiaz> nijaba: I'm not sure it'd be as techinical as a how-to
[23:56:04] <mathiaz> nealmcb: I think so
[23:56:14] <nijaba> mathiaz: It needs to be two fold
[23:56:45] <mathiaz> nijaba: agreed - that's why we should discuss what would the puprose of a white paper
[23:57:04] <mathiaz> nijaba: I'd defer that to the mailing-list or a uds topic
[23:57:10] <mathiaz> nijaba: but I think the idea is a good one
[23:57:14] <mathiaz> [TOPIC] #
[23:57:14] <mathiaz> Agree on next meeting date and time.
[23:57:16] <nealmcb> I agree that a good set of white papers would be great
[23:57:21] <nijaba> ok, so I'll try to push it forward
[23:57:25] <mathiaz> [TOPIC] Agree on next meeting date and time.
[23:57:31] <owh> faulkes-: I once wrote an article in 2003 about how to get Fortune 1000 companies to adopt Linux: http://itmaze.com.au/articles/cio/
[23:57:41] <owh> mathiaz: How did the bug for elmo go that we discussed last week? Bug 189616
[23:57:41] <mathiaz> Same place, same time, next week ?
[23:57:44] <ubotu> Launchpad bug 189616 in dovecot "connection problems under load with hardy dovecot" [Undecided,Incomplete] https://launchpad.net/bugs/189616
[23:57:58] <mathiaz> owh: I'm still looking into it - haven't reproduced yet
[23:58:15] <owh> mathiaz: Was elmo able to provide more information?
[23:58:38] <mathiaz> owh: not a lot
[23:58:48] <mathiaz> owh: let's discuss that in #u-server
[23:58:54] <owh> Yup.
[23:59:07] <mathiaz> so next week, same place, same time ?
[23:59:20] <sommer> +1
[23:59:27] <owh> cjwatson: FYI: The other applications I was looking at for the man page to html were: man2html, dwww, info2www
[23:59:31] <owh> mathiaz: +1
[23:59:44] <mathiaz> alright - see ya next week and happy rc testing !
[23:59:46] <mathiaz> #endmeeting

MeetingLogs/Server/20080416 (last edited 2008-08-06 16:32:29 by localhost)