IRC chat logs for #ltsp on irc.freenode.net (webchat)


Channel log from 20 February 2012   (all times are UTC)

00:23monteslu has left IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net, Read error: Operation timed out)
00:35monteslu has joined IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net)
01:09killermike has left IRC (killermike!~killermik@2.26.113.196, Remote host closed the connection)
01:17loather has joined IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net)
01:45alexqwesa has left IRC (alexqwesa!~alex@109.172.15.11, Quit: Хана X'ам !!!)
02:20staffencasa has joined IRC (staffencasa!~staffenca@128-193-148-241.oregonstate.edu)
02:27adrianorg_ has joined IRC (adrianorg_!~adrianorg@186.213.157.3)
02:27adrianorg has left IRC (adrianorg!~adrianorg@187.115.105.16, Read error: Connection reset by peer)
02:53adrianorg_ has left IRC (adrianorg_!~adrianorg@186.213.157.3, Ping timeout: 252 seconds)
03:56risca has left IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca, Quit: Lämnar)
03:56risca has joined IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca)
04:50map7 has joined IRC (map7!~map7@teksup41.lnk.telstra.net)
04:50
<map7>
How do I make an overlay for LTSP clients?
04:51
And should I use overlayfs, aufs or unionfs?
04:51
<vagrantc>
what are you trying to do?
04:51
i.e. why do you want an overlay?
04:52
<map7>
I have an old LTSP server for mythtv clients and it was built using the mythbuntu flags on LTSP which gave us an overlay
04:52
This allowed us to use this to customise each fat client to our needs and leave it in the NFS shared overlay
04:52
It also allowed us to use the same user for every client without any app conflicting
04:53
<vagrantc>
could see that possibly being useful for LTSP upstream... we should maybe port it.
04:53
but currently, there isn't that sort of thing integrated into LTSP...
04:54
<map7>
It was a great feature, but over the last two years it has not been maintained and now the 'mythbuntu' switches are unusable
04:54
<vagrantc>
i think most versions in the wild use aufs for what simple run-time overlays we do, but ubuntu is switching to overlayfs ... maybe really ancient versions used unionfs
04:54map7work has left IRC (map7work!~map7@teksup41.lnk.telstra.net, Ping timeout: 276 seconds)
04:55
<map7>
I know the mythbuntu flags in ltsp-build-client setup 'aufs'
04:55
<vagrantc>
i think ubuntu's current development release droppped support for aufs.
04:55
<map7>
And I've also read that overlayfs is newer and they way to go, but thought I would check
04:56
If I wanted to build multiple fat-clients to all start the same application (XBMC for instance) should I create a separate user for each thin client or can I use the same user login for all?
04:57
ie: will there be conflicts and apps crashing if I have the same user logged in on more than one machine?
04:57
<vagrantc>
sounds like you're just left to experimenting at this point, but i'd be interested in integrating some sort of feature like what you're taking about
04:57
map7: if you're using a shared homedir, that will be a problem.
04:57
map7: but if they're fat clients each with their own homedir, it should be fine.
04:58
by default, fat clients use a shared homedir mounted using sshfs (with options to mount over NFS)
04:58
though you might be able to implement some sort of overlay on top of that at login time
04:59
anyways, gotta run. good luck, and be interesting to hear about what you come up with!
04:59
<map7>
ok catch you later
05:01vagrantc has left IRC (vagrantc!~vagrant@75-150-46-245-Oregon.hfc.comcastbusiness.net, Quit: leaving)
06:12Parker955 is now known as Parker955_Away
06:30alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
06:34NeonLicht has left IRC (NeonLicht!~NeonLicht@darwin.ugr.es, Ping timeout: 260 seconds)
06:35NeonLicht has joined IRC (NeonLicht!~NeonLicht@darwin.ugr.es)
06:40macele has joined IRC (macele!~macele@cpe-24-167-29-26.rgv.res.rr.com)
06:56NeonLicht has left IRC (NeonLicht!~NeonLicht@darwin.ugr.es, Ping timeout: 244 seconds)
06:56NeonLicht has joined IRC (NeonLicht!~NeonLicht@darwin.ugr.es)
07:05macele is now known as macele_deceased
07:28staffencasa has left IRC (staffencasa!~staffenca@128-193-148-241.oregonstate.edu, Ping timeout: 244 seconds)
08:02loather has left IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net, Quit: This computer has gone to sleep)
08:04skumlesen has joined IRC (skumlesen!~skumlesen@87-104-113-212-dynamic-customer.profibernet.dk)
08:16sep has left IRC (sep!~sep@40.211.jostedal.no, Read error: No route to host)
08:18sep has joined IRC (sep!~sep@40.211.jostedal.no)
08:45Steve_The_Pirate has joined IRC (Steve_The_Pirate!~Gary@ip-80-238-8-128.bskyb.com)
08:49dobber has joined IRC (dobber!~dobber@213.169.45.222)
09:52alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Remote host closed the connection)
10:04Steve_The_Pirate has left IRC (Steve_The_Pirate!~Gary@ip-80-238-8-128.bskyb.com, *.net *.split)
10:04sunson_ has left IRC (sunson_!~suraj@li290-74.members.linode.com, *.net *.split)
10:04sunson has joined IRC (sunson!~suraj@li290-74.members.linode.com)
10:10khildin has joined IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be)
10:17Steve_The_Pirate has joined IRC (Steve_The_Pirate!~Gary@ip-80-238-8-128.bskyb.com)
10:44gentgeen__ has left IRC (gentgeen__!~kevin@c-98-236-71-64.hsd1.pa.comcast.net, Remote host closed the connection)
10:44alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
11:20killermike has joined IRC (killermike!~killermik@2.26.103.147)
11:32adrianorg_ has joined IRC (adrianorg_!~adrianorg@186.213.157.3)
11:43artista_frustrad has joined IRC (artista_frustrad!~fernando@200.247.43.2)
11:52Trixboxer has joined IRC (Trixboxer!~Trixboxer@115.124.115.71)
12:22monteslu has left IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net, Read error: Operation timed out)
12:31skumlesen has left IRC (skumlesen!~skumlesen@87-104-113-212-dynamic-customer.profibernet.dk, Quit: bbl - work work work)
12:35monteslu has joined IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net)
12:44alexqwesa has joined IRC (alexqwesa!~alex@alexo-veto.broker.freenet6.net)
12:57cyberorg has left IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg, Remote host closed the connection)
13:00cyberorg has joined IRC (cyberorg!~cyberorg@opensuse/member/Cyberorg)
13:09brunolambert has joined IRC (brunolambert!bruno@nat/revolutionlinux/x-xslibmcvkogpsxpv)
13:16alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
13:27Da-Geek has joined IRC (Da-Geek!~Da-Geek@94.236.7.190)
13:38mgariepy has joined IRC (mgariepy!mgariepy@ubuntu/member/mgariepy)
13:53
<mgariepy>
good morning everyone
13:54
<elias_a>
mgariepy: Good afternoon :)
14:15
<Hyperbyte>
Good afternoon! ;-)
14:17
<elias_a>
Hyperbyte: So you're not walking upside down like mgariepy is doing on the other side of the globe? :P
14:17
<Hyperbyte>
elias_a, no, I'm from the Netherlands. :-)
14:18
<elias_a>
Hyperbyte: I thought so after looking at your realname ;-)
14:18dead_inside has joined IRC (dead_inside!~dead_insi@76.75.3.174)
14:21
<Hyperbyte>
How are you today? :)
14:28
<elias_a>
Hyperbyte: Fine, thanks!
14:29
Actually really good considering that I have being thinking how the use of FLOSS could be easier for Windoze admins....
14:29
No wonder I have a starting headache... :P
14:30
Hyperbyte: Did the waterways freeze for the skating competition?
14:30
The narrow channels - that's what I mean.
14:37bobby_C has joined IRC (bobby_C!~bobby@85-124-22-227.teleworker.xdsl-line.inode.at)
14:40
<||cw>
elias_a: how is it hard? I've been using open source on windows for many years. though I generally let someone else compile it for me :)
14:42
<elias_a>
||cw: I am talking about networks with centralized maintenance and hundreds or even thousands of workstations.
14:44
||cw: They want easy also when it comes to controlling policies.
14:44
<||cw>
so like AD and windows update server?
14:44
I'm just not sure there's really a market for that
14:44
<elias_a>
They are interested in those things as they still have not seen the light of LTSP :P
14:45
||cw: Well - imho there is in some things: distributing sw like Firefox or Scribus...
14:45
But that's a bit OT.
14:46
<||cw>
LTSP is nice, but if you need windows software it's not always a usable solution. as the freerdp project progresses and rdp graphics performance improves, ltsp will make a very nice VDI system
14:46
there's MSI's for firefox, and group policy templates
14:46
not sure abut scribus, but no reason it couldn't be done
14:46
<elias_a>
||cw: Yep - but not by the project itself. It is 3rd party.
14:47
<||cw>
has the 3rd party submited the patches?
14:49
I do agree it could be easier, but someone has to do the work, and it's got to quality work submitted and vetted, and with users putting pressure on including the features
14:50
<elias_a>
||cw: In don't know. I noticed it today: http://www.frontmotion.com/Firefox/
14:50
<||cw>
as long as too few admins are pushing FF to include it, they won't.
14:53Parker955_Away is now known as Parker955
14:53
<||cw>
here's one that does thunderbird too http://www.zettaserve.com/products/free-msis/
15:05
<Hyperbyte>
elias_a, the waterways did freeze for lots of skating competitions, but not -the- skating competition of the Netherlands... lots and lots of people participate in that, and it takes a few days to organize... we had maybe three or four days of +10cm ice...
15:07Parker955 is now known as Parker955_Away
15:08
<elias_a>
Hyperbyte: Sad that the big event could not be organized. Would have been fun to watch!
15:11monteslu has left IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net, Read error: Operation timed out)
15:15Gremble has joined IRC (Gremble!~Ben@cpc10-aztw24-2-0-cust114.aztw.cable.virginmedia.com)
15:15
<Gremble>
alkisg
15:15
oh, not here
15:16
hoping to say that using LDM_USER_ALLOW works really well
15:16
I'll call back later
15:16Gremble has left IRC (Gremble!~Ben@cpc10-aztw24-2-0-cust114.aztw.cable.virginmedia.com, Client Quit)
15:19bakytn has joined IRC (bakytn!bakytn@80.72.185.225)
15:19
<Hyperbyte>
elias_a, to be honest, I don't care much for it. :)
15:23monteslu has joined IRC (monteslu!~monteslu@ip68-109-174-213.ph.ph.cox.net)
15:27komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:28komunista has left IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk, Remote host closed the connection)
15:28komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:30komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:32alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
15:35komunista has left IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk, Remote host closed the connection)
15:36komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:39komunista has left IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk, Client Quit)
15:39skumlesen has joined IRC (skumlesen!~skumlesen@mail.kia.dk)
15:41komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:43komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
15:45alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Quit: Leaving.)
15:46komunista has left IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
16:03adrianorg_ is now known as adrianorg
16:06bakytn has left IRC (bakytn!bakytn@80.72.185.225, Remote host closed the connection)
16:06bakytn has joined IRC (bakytn!~bakytn@80.72.185.225)
16:07Steve_The_Pirate has left IRC (Steve_The_Pirate!~Gary@ip-80-238-8-128.bskyb.com, Quit: Leaving)
16:12vagrantc has joined IRC (vagrantc!~vagrant@c-76-115-60-19.hsd1.or.comcast.net)
16:18spectra has left IRC (spectra!~spectra@debian/developer/spectra, Ping timeout: 245 seconds)
16:19spectra has joined IRC (spectra!~spectra@debian/developer/spectra)
16:21bakytn has left IRC (bakytn!~bakytn@80.72.185.225, Remote host closed the connection)
16:22bakytn has joined IRC (bakytn!~bakytn@80.72.185.225)
16:24komunista has joined IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk)
16:27
<vagrantc>
hmmm... seems like configure_localdev is called both from ltsp-init.d and from udev ...
16:27staffencasa has joined IRC (staffencasa!~staffenca@128-193-148-241.oregonstate.edu)
16:27
<vagrantc>
ah, there's a FIXME for it...
16:33dobber has left IRC (dobber!~dobber@213.169.45.222, Remote host closed the connection)
16:35bakytn has left IRC (bakytn!~bakytn@80.72.185.225, Remote host closed the connection)
16:35bakytn has joined IRC (bakytn!~bakytn@80.72.185.225)
16:37shawnp0wers has joined IRC (shawnp0wers!~spowers@71-13-74-18.static.aldl.mi.charter.com)
16:37
<bakytn>
guys! is there any workarounds for not using ltsp-update-image everytime I change something with NBD. I could use NFS but AFAIK it's very very slow.
16:37shawnp0wers has left IRC (shawnp0wers!~spowers@71-13-74-18.static.aldl.mi.charter.com, Changing host)
16:37shawnp0wers has joined IRC (shawnp0wers!~spowers@linuxjournal/staff/shawnp0wers)
16:38
<bakytn>
I am adding printers and having to ltsp-update-image all the time yet. thinking about writing some RC script. But may be there more elegant/faster solution?
16:38
<vagrantc>
bakytn: you could use a writeable filesystem like ext2 instead of squashfs
16:39
<bakytn>
vagrantc, cool! how? and is there any drawbacks?
16:40
<highvoltage>
bakytn: http://www.thegeekstuff.com/2009/02/nbd-tutorial-network-block-device-jumpstart-guide/ shows how to make a loopback filesystem for squashfs
16:40
<vagrantc>
bakytn: sqaushfs is compressed, so you transfer less data over the wire with squashfs
16:40
<bakytn>
vagrantc, with this, I could install device drivers then! when it wasnot possible with my current setup
16:40
<highvoltage>
(it's not ltsp specific but it gives you a change to learn how it works :) )
16:41
<bakytn>
highvoltage, thanks!
16:41
vagrantc, thanks!
16:41macele_deceased is now known as macele
16:41
<vagrantc>
bakytn: although once you change over, you won't want to run ltsp-update-image anymore, as it will obliterate your existing image.
16:42
<knipwim>
vagrantc: debian uses aufs over nfs/nbd as well?
16:42
<vagrantc>
knipwim: as of later today, yes.
16:43
<knipwim>
is the nfs mounted in a new way to support for aufs?
16:43
i'm getting /rofs is not a block device
16:43
<vagrantc>
knipwim: nope, there's a hook in the latter part of initramfs-tools that mount -o move's stuff around
16:44
knipwim: ltsp-trunk/client/initramfs/scripts/init-bottom/ltsp
16:44bakytn has left IRC (bakytn!~bakytn@80.72.185.225, Remote host closed the connection)
16:44
<knipwim>
yeah, the init-bottom/ltsp
16:44
id :)
16:44bakytn has joined IRC (bakytn!~bakytn@80.72.185.225)
16:44
<knipwim>
i'm using the exact part of the script, but then in dracut
16:45
perhaps i should ask there
16:46
at least nfs is not causing the problem, dracut and aufs left
16:54Da-Geek has left IRC (Da-Geek!~Da-Geek@94.236.7.190, Quit: Leaving)
17:01
<vagrantc>
!seen alkisg
17:01
<ltsp`>
vagrantc: alkisg was last seen in #ltsp 19 hours, 21 minutes, and 12 seconds ago: <alkisg> It'd be nice to have them in the next Debian/Ubuntu releases...
17:01
<vagrantc>
!logs
17:01
<ltsp`>
vagrantc: I do not know about 'logs', but I do know about these similar topics: 'ldm', 'leet', 'ltsp'
17:02
<knipwim>
vagrantc: http://irclogs.ltsp.org/
17:02
<vagrantc>
wow, irclogs.ltsp.org is way cooler than the last logs implementations i've seen!
17:02bakytn has left IRC (bakytn!~bakytn@80.72.185.225, Remote host closed the connection)
17:02bakytn has joined IRC (bakytn!~bakytn@80.72.185.225)
17:03
<knipwim>
you can actually search
17:04
!learn logs http://irclogs.ltsp.org/
17:04
<ltsp`>
knipwim: (learn [<channel>] <key> as <value>) -- Associates <key> with <value>. <channel> is only necessary if the message isn't sent on the channel itself. The word 'as' is necessary to separate the key from the value. It can be changed to another word via the learnSeparator registry value.
17:04
<knipwim>
!learn logs as http://irclogs.ltsp.org/
17:04
<ltsp`>
knipwim: The operation succeeded.
17:12
<Hyperbyte>
vagrantc, thanks. :-D
17:13
<vagrantc>
Hyperbyte: i get to blame you for the improvemment! good show!
17:19
<Hyperbyte>
:)
17:22
<knipwim>
Hyperbyte!
17:22
how's it going, settling in nicely?
17:23alkisg has joined IRC (alkisg!554b8f18@gateway/web/freenode/ip.85.75.143.24)
17:25alkisg has left IRC (alkisg!554b8f18@gateway/web/freenode/ip.85.75.143.24, Client Quit)
17:40
<Hyperbyte>
Hey knipwim... it's going better. I was a little bit overworked after the move, but I'm slowly settling again.
17:41
Hopefully I'll be getting an IT colleague soon, because the workload since the new office + extra company that moved in with us is becoming a bit insane.
17:41alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg)
17:51brunolambert has left IRC (brunolambert!bruno@nat/revolutionlinux/x-xslibmcvkogpsxpv, Quit: brunolambert)
17:52khildin has left IRC (khildin!~khildin@ip-80-236-223-18.dsl.scarlet.be, Quit: I'm gone, bye bye)
17:54
<knipwim>
Hyperbyte: and maintance for LTSP was supposed to be easy ;)
17:55
few extra clients, no problem
17:55
but seriously, the extra company didn't bring it's own IT staff?
17:56
and still expect you to do twice the work in the same time
17:59
<||cw>
well, anytime you move or integrate some extra short term IT work should be expected until everyone is merged and acclimated to the new systems
18:03
<knipwim>
true, lots of user support
18:16loather has joined IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net)
18:35
<alkisg>
vagrantc, stgraber: I'm trying to clean up a bit the LDM_SESSION-related bugs. What do we want LDM_SESSION to contain, the value "gnome-classic", i.e. the same that .dmrc contains, or the value "gnome-session --session=gnome-classic" , i.e. the Exec line of the gnome-classic.desktop line?
18:35
I believe that LDM_SESSION=gnome-classic makes more sense, but I don't know how people have used it so far...
18:37
<stgraber>
alkisg: ideally it should be a xsession file name as found in /usr/share/xsessions/
18:37alexqwesa has left IRC (alexqwesa!~alex@alexo-veto.broker.freenet6.net, Ping timeout: 245 seconds)
18:37
<stgraber>
which should match what you get in a .dmrc on a regular desktop machine (and hopefully with ldm too)
18:37
<alkisg>
Cool. OK, I'll do some changes, this would fix this bug as well: https://bugs.launchpad.net/ltsp/+bug/820417
18:37
<vagrantc>
with the latest version, SCREEN_NN isn't working quite right.
18:38
oh, might be my mistake
18:40
ah, i was editing lts.conf while it was booting, and it decided to use the old version... so my mistake, it's working fine :)
18:43
alkisg: so, we've got predictible names for the nbd swap stuff...
18:43alexqwesa has joined IRC (alexqwesa!~alex@109.172.15.11)
18:44
<alkisg>
vagrantc: it'd be nice if we had a dir /var owned by nobody:nogroup...
18:44
(you're talking about static names in /tmp, right?)
18:45
<vagrantc>
alkisg: yes, skince anyone can write to /tmp, it's generally bad form to have a predictible name there...
18:45
i guess it's *less* of a problem using a sub-dir
18:45
<alkisg>
Right. So, we'd either run nbd-server as root, or have a service create a dir in /var for us, with owner=the nbd-server userid
18:46
...and I don't know what service would that be, we don't have a server-side ltsp service, do we?
18:46
<vagrantc>
would also be nice to be able to use a non-nobody user
18:46
since other services may be using the nobody user
18:46
<alkisg>
$ grep ^user /etc/nbd-server/config
18:46
user = nobody
18:46
Whatever that uses
18:47
I.e. the service should read that file before setting the dir owner
18:47
<vagrantc>
it uses the user "nobody", but would be more ideal to have a separate user.
18:47
<alkisg>
In Ubuntu I think we have an nbd user
18:47
<vagrantc>
this is a separate issue
18:47Parker955_Away is now known as Parker955
18:47
<alkisg>
I thought stgraber would copy the nbd example config file, along with the nbd user stanza...
18:48
....but I haven't re-installed ltsp-server in my pc, not sure if I got it wrong here
18:48
<vagrantc>
yeah, appears to be an nbd user on debian too ... although i don't know if we'd be abusing that
18:48
<alkisg>
Isn't that exactly why it's there?
18:48
<vagrantc>
don't know why it's there
18:48
<alkisg>
cat /usr/share/nbd-server/nbd-server.conf.tmpl
18:49
<vagrantc>
i could see that being a good rreason for it to be there, but i wouldn't want to make assumptions :)
18:49
<alkisg>
It's part of the example configuration file that nbd ships
18:49
So, why not use that, it suits us fine
18:49
<vagrantc>
yeah, ok.
18:49
works for me :)
18:50
<alkisg>
Let me see what our postinst contains...
18:50
Hmm it needs fixing
18:51
<vagrantc>
ideally we'd have the nbd swap stuff writeable by the nbd user, but the other (read-only) stuff writeable only by root.
18:51
<alkisg>
stgraber: in ltsps-server.postinst, we want to copy /usr/share/nbd-server/nbd-server.conf.tmpl into /etc/nbd-server/config, if the target doesn't exist, and we don't want port 9572 in inetd anymore
18:52
<vagrantc>
i'm still getting the /var/cache/apt problem :(
18:52* alkisg thinks it'd be nice if debian+ubuntu had less packaging diff in ltsp
18:52
<vagrantc>
worst case, i guess ii remove the files on boot
18:52
<alkisg>
vagrantc: did you copy some stuff from init-ltsp.d/Ubuntu to Debian or to common?
18:52
<vagrantc>
alkisg: yeah, we've really diverged a lot there.
18:52
alkisg: i just made it a symlink
18:53
<alkisg>
All of those files?
18:53
<vagrantc>
ln -s Ubuntu Debian
18:53
<alkisg>
Hm. I wonder why then..
18:53
<vagrantc>
haven't pushed it in my packages yet, but it hasn't seemed to cause any problems
18:53
i'm guessing debian names something different that runs some annoying service
18:54
<alkisg>
That, or cron jobs
18:54
<vagrantc>
also wondered if i shouldn't just purge cron.daily, cron.weekly and cron.monthly at boot.
18:54
<stgraber>
alkisg: I thought ltsp-update-image was supposed to do the copying
18:54
alkisg: but yeah, easy enough to add the cp in the postinst ;)
18:54
<vagrantc>
it would also be really useful to get the nbd-config setting stuff out of ltsp-update-image and into a separate file ...
18:55
or i should port ltsp-update-image to work with other NBD image formats...
18:55
<alkisg>
Other formats? Meaning?
18:55
<vagrantc>
non-squashfs images
18:55rthomson has joined IRC (rthomson!~rthomson@mars.pet.ubc.ca)
18:55
<alkisg>
What's to update there?
18:56
<vagrantc>
the nbd-server configs, the pxelinux boot options
18:56
ltsp-update-image --no-squashfs
18:56
that might do the trick for me.
18:57
<alkisg>
I wouldn't call that "image" anymore though
18:57
ltsp-export-image, ok
18:57
<vagrantc>
sure.
18:58
ltsp-update-image is doing more than updating the image ... i'd like to get the non-image functionality either split out or at least callable with commandline options
18:58
see Debian/035-create-fs-image
18:59
<alkisg>
vagrantc: https://help.ubuntu.com/community/HowToSetupLTSPDevelEnvironment#Using_an_NBD_file-based_.22partition.22
19:00
btrfs works better because it's compressed ;)
19:00
Yup I know what you mean, I'd also like some hook for vdi2squashfs
19:00
<vagrantc>
sure, ltsp-update-image shouldn't be tied to only work with a specific filesystem
19:01
it could fix that "Remember not to run ltsp-update-image afterwards." bug.
19:01
which has plauged debian for some time
19:02
<alkisg>
Indeed
19:02
<vagrantc>
and it seems kind of wrong tto ship ltsp-update-image.conf with "exit 0" or some such
19:02
very curious to experiment with btrfs images!
19:02
it supports writeable but compressed images?
19:03
<alkisg>
They're about twice the size of squashfs
19:03
I've used them for quick testing while developing
19:03
<vagrantc>
looks like we've added an iproute dependency to ltsp-client
19:04
<alkisg>
They're writeable too, I tried that without ltsp. But we could use it to store per-client image diffs.
19:04
<vagrantc>
oh, i was just meaning running apt-get upgrade without rebuilding the image
19:05
<alkisg>
Yeah I even made a little `btrfs-chroot` script, which stops nbd-server, loop mounts the image, then runs ltsp-chroot -m, then restarts nbd-server
19:05
Very handy
19:05
Didn't we always use the `ip` command?
19:06
vagrantc: if you have a minute about a piuparts failure about epoptes: http://piuparts.debian.org/sid/fail/epoptes_0.4.2-1.log
19:06
1m0.8s ERROR: FAIL: Package purging left files on system:
19:06
/etc/epoptes not owned
19:06
/root/.rnd not owned
19:07
I can easily fix the /etc/epoptes dir, ok
19:07
<vagrantc>
alkisg: maybe we did, but i just noticed it failing when ip was missing.
19:07
<alkisg>
But /root/.rnd is created by the openssl call
19:07
<vagrantc>
alkisg: i think we used it server-side, but not client-side before.
19:07
alkisg: where's the /root/.rnd coming from?
19:07
ah.
19:08
<alkisg>
When we create the server certificate at postisnt
19:08
http://www.openssl.org/support/faq.html#USER1
19:08
<vagrantc>
alkisg: only thing i can think of is to call openssl with a different homedir or something that we delete
19:09
<alkisg>
Right, but I think that's worse than a piupart error... :-/
19:09
..but ok whatever debian maintainers suggest :)
19:09
<vagrantc>
or set RANDFILE
19:10
or the -rand option?
19:10brunolambert has joined IRC (brunolambert!bruno@nat/revolutionlinux/x-mybpgqekwjqxticc)
19:12
<vagrantc>
alkisg: i think setting export RANDFILE=$(mktep) might be a sane way to go.
19:12
<alkisg>
Setting RANDFILE doesn't work for me
19:12
<vagrantc>
ok, so that doesn't work.
19:12
:)
19:12
<alkisg>
I think it can be used as the source, but not as the "output" of the random generator state
19:14
So yeah setting HOME=$(mktemp -d) was the only way I could think, but it sounds like a hack to me
19:14floorduh has joined IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
19:14
<vagrantc>
or just ignore the error.
19:15
<alkisg>
:)
19:15
<vagrantc>
alkisg: could also ask about it in #debian-devel
19:15
or #debian-mentors
19:15
<alkisg>
Ty, let me try that...
19:15
<vagrantc>
(on irc.oftc.net)
19:25
one advantage of doing more in the initramfs is that you don't have to run ltsp-update-image to deploy changes that only affect the initramfs...
19:25
although we weren't doing that with initramfs-scripts.d, since most of the code lived in the chroot/image
19:26
i cannot figure out a rhyme or reason for the /var/cache/apt stuff ...
19:27
sometimes it doesn't create the files, sometimes it does.
19:27
if they were smaller files, i wouldn't sweat it... but 30MB+ of ram is kind of wasteful.
19:28
cdpinger is getting started twice :(
19:30
ah, it's called both by udev and by init-ltsp.d/*localdev
19:31
<alkisg>
vagrantc, stgraber, would you agree with making ltsp-client installable on regular machines? https://bugs.launchpad.net/ltsp/+bug/936810
19:31
There are some packaging tasks there that I cannot do :-/
19:31
Part of those is that thing about udev ^ :)
19:32
<vagrantc>
makes me nervous, but we're so close.
19:32
alkisg: also, was wondering if a case statement would work better for ltsp-client-core than the grep?
19:34skumlesen has left IRC (skumlesen!~skumlesen@mail.kia.dk, Ping timeout: 240 seconds)
19:35
<alkisg>
vagrantc: for speed? When we fix the whole configuration system, it should just check if LTSP_BOOT is active...
19:35
<vagrantc>
alkisg: yeah, speed reasons... also code consistancy
19:35
although our other code calls cat
19:36
although cat is about 1/3rd the size of grep :)
19:36
and calls less libraries
19:36
<alkisg>
The other code also wants to do different stuff based on if it's "ltsp" or "init=..."
19:37
<vagrantc>
yes, but that's not a big deal
19:37
i think the other code is maybe more readable... dunno really... was just a thought
19:37
<alkisg>
Well, that other code should save the result in our config, and ltsp-client-core should source our config, not check again the command line
19:37
But we're not there yet, the config stuff isn't working
19:37
<vagrantc>
it isn't?
19:37
<alkisg>
No, it's a mess
19:37
<vagrantc>
ah, you mean the ltsp_config insanity?
19:37
<alkisg>
Yeah
19:38
<vagrantc>
well, that we have both init-ltsp.d and ltsp_config.d ... both with overlapping functionality and purpose.
19:38* vagrantc suspects tagging 5.3 was a bit premature, if we really want to tackle that
19:38
<vagrantc>
i guess it's the new way forward...
19:39
<alkisg>
stgraber: join us on that please ^
19:39
I could try to help with ltsp_config.d
19:39
And it's broken now
19:39
But, can we do that on the precise cycle?
19:39
Or should we quickly fix the big bugs, and leave the proper solution for later on?
19:40
<vagrantc>
i don't see it being completely broken, i see it being a little messy.
19:40
<alkisg>
E.g. now all that "set_lts_var" stuff calls sed 2 times for each lts var
19:40
That's very inefficient for a caching mechanism that's supposed to speed up things
19:40
<vagrantc>
this is probably true.
19:41
i do recall the deletion mechanism not really being necessary
19:41
i.e. if we just keep adding to it, it actually works fine.
19:41
<alkisg>
So we should probably start by defining points that we get the configuration, like "initramfs", "init", "login", "logout", like in -cluster,
19:41
<vagrantc>
it's cheaper to read a variable setting multiple times than to delete the variable when we reset it ... i remember this conversation at the BTS
19:41
<alkisg>
and if we want deletions, then at the start of those points we can reevaluate all the configuration
19:42
Also, the values calculated by initramfs and init-ltsp.d are static, those events only happen once
19:42
They should be in separate cache files
19:43
<vagrantc>
isn't that what ltsp_cache vs. ltsp_cache_env was for?
19:43
<alkisg>
ltsp_config_env, yeah
19:44
So we shouldn't be using set_lts_var in init-ltsp.d, and we should use it later on? It's messy
19:44
It'd be better if that was handled by the "config backend"
19:44
<vagrantc>
i don't really have a good handle on the situation
19:44
<alkisg>
Example, if "set_lts_var" knew the "boot phase", so that it knows where to store it
19:45
<vagrantc>
the problem is folks want to reset some variables?
19:45
<alkisg>
Neither do I, but we did talk with Gadi a lot about a new configuration mechanism
19:45
And implemented part of it, outside ltsp
19:45
<vagrantc>
i mean, getltscfg -a wasn't so bad ... and then we went crazy
19:46
<highvoltage>
vagrantc: howdy! I'm looking at LTSP theming a bit (well, at https://bugs.launchpad.net/ltsp/+bug/816150 in particular but that's not entirely relevant)
19:46
<alkisg>
True, I think we've lost track of why there were design changes
19:46
<vagrantc>
that said, what we have now does actually work, as far as i can tell.
19:46
<highvoltage>
vagrantc: I can't remember where your last suggestions on the default was, did you say we should switch gtk theme to one with less dependencies?
19:47
<vagrantc>
highvoltage: for ldm-themes?
19:47
<highvoltage>
the the default ltsp theme that comes with ltsp
19:47skumlesen has joined IRC (skumlesen!~skumlesen@mail.kia.dk)
19:47
<vagrantc>
upstream defaults to clearlooks, and debian now follows suit.
19:48
i would have lliked to try murrine, but the debian maintainer is obstinant about a needless recommends.
19:49
<highvoltage>
pl
19:49
*ok
19:49
<alkisg>
ij, a little more to the left
19:49
<vagrantc>
http://bugs.debian.org/623783
19:50
supposedly gtk3 makes it a moot point?
19:51
<highvoltage>
not sure if it's entirely moot
19:51
but... it's probably possible to move to a gtk3 theme that's smaller
19:51
can I look into it?
19:51
<vagrantc>
right, that's the goal...
19:52
in debian, clearlooks is bundled with a bunch of other engines in gtk2-engines, which is unfortunate, but i got weary of maintaining a diff with upstream, and there's no better alternative at the moment.
19:53
highvoltage: might be good to do the samme with ldm-themes in debian.
19:53
<highvoltage>
yeah
19:54
<vagrantc>
then i'd consider installing it by default :)
19:54
<highvoltage>
:)
19:55
<vagrantc>
i'll make sure to use an alternate recommends, though
19:55
recommends: ldm-themes | ldm-theme
19:55
and then other packages, should they ever exist, can Provide: ldm-theme
19:56
<highvoltage>
gtk3-engines-unico doesn't recommend on any themes and it's less than 30K
19:56
<vagrantc>
we don't yet support gtk3, as far as i know.
19:57
though that sounds promising :)
19:57
<highvoltage>
stgraber: I was under the impression that you migrated ldm to gtk3, am I wrong?
19:58
<vagrantc>
last i heard it was mostly done, but not quite done.
19:58
and not pushed to ldm-trunk
19:59
<alkisg>
Ouch, ldminfod returns "gnome-session --session=gnome-classic", not just "gnome-classic". We'd need to modify that too if LDM_SESSION is supposed to get the session file name instead of the Exec line.
19:59
<stgraber>
right, it's currently a separate branch that's still missing some bits
19:59
<alkisg>
But that will break connections from older chroots
19:59
<stgraber>
I'm happy to push it to LP if someone else in ~ltsp-upstream has time to finish the work
20:00
<vagrantc>
alkisg: can always add another ldm-session-desktop-file or something
20:01
<stgraber>
pushed lp:~ltsp-upstream/ltsp/ldm-gtk3
20:02
<vagrantc>
currently there's session: and session-with-name: could introduce session-desktop-file-session:
20:02
<stgraber>
(and rebased on current ldm-trunk)
20:03
<vagrantc>
alkisg: although it would have to fetch the corresponding .desktop file from the server, no?
20:03
<highvoltage>
vagrantc: what about gtk2-engines-cleanice? It's also under 30k and has no theme dependencies
20:03
<alkisg>
vagrantc: we do that anyway, in dmrc-processing
20:03
<vagrantc>
alkisg: ok.
20:04
highvoltage: is it maintained?
20:04gado has joined IRC (gado!5f0f6c5d@gateway/web/freenode/ip.95.15.108.93)
20:05
<alkisg>
So... the main decision is that we do want LDM_SESSION to correspond to the desktop file name. I think I can do it by only changing ldm and dmrc-processing and run-x-session, without changing ldminfod
20:06
<vagrantc>
alkisg: with backwards compatibility?
20:06
<alkisg>
I.e., ldm will either set LDM_SESSION or LDM_SESSION_EXEC, and dmrc will then set both of them, and write LDM_SESSION to .dmrc, and run-x-session will use LDM_SESSION_EXEC
20:06
vagrantc: yes
20:06
LDM_SESSION will come from lts.conf, LDM_SESSION_EXEC from the greeter
20:07
<vagrantc>
alkisg: seems worth doing to me.
20:07
<alkisg>
And when we reimplement ldminfod to substitute lts.conf too, we can get rid of LDM_SESSION_EXEC
20:07
OK, on to it...
20:07
<vagrantc>
as long as it doesn't add ridiculous overhead or code.
20:07
<alkisg>
No, only a few lines in dmrc-processing
20:08
...which needs a bit of a rewrite anyway
20:08
The only breakage in backwards compatibility will be this:
20:08
!gnome-classic
20:08
<ltsp`>
alkisg: gnome-classic: To select gnome-classic as your default session, put this line in your lts.conf: LDM_SESSION="gnome-session --session=gnome-classic"
20:09
<alkisg>
...from now on, it'll be LDM_SESSION=gnome-classic
20:09
<vagrantc>
alkisg: that doesn't sound like it doesn't break backwards compatibility
20:10
<alkisg>
Wait, now LDM_SESSION has two meanings
20:11
<vagrantc>
highvoltage: dunno if other distros ship cleanice, but that's another consideration
20:11
<alkisg>
It's either read from .dmrc, or provided by lts.conf, or selected in gtk greeter
20:11
<vagrantc>
highvoltage: clearlooks has been around for ages...
20:11
<alkisg>
The first one, .dmrc, has different contents than the greeter one
20:11
<vagrantc>
so it's broken?
20:11
<alkisg>
Yes
20:11
https://bugs.launchpad.net/ltsp/+bug/820417
20:12
And we need to decide what the middle one, lts.conf, means
20:12
<vagrantc>
originally, it had nothing to do with .desktop files.
20:12
<alkisg>
Does it mean the same as .dmrc, or the same as our greeter?
20:12
<vagrantc>
i think someone tried to change it to "work" with desktop files, and didn't finish
20:12
<highvoltage>
vagrantc: it's last upload was mid 2010, I guess it doesn't have any further updates because the last few uploads closed all of its bugs (and there are no current ones filed against it)
20:13
<alkisg>
I do think that people expect LDM_SESSION to be the same as .dmrc session= lines
20:13
<vagrantc>
highvoltage: no new upstream releases...
20:13
<alkisg>
We could allow both, but we'd need another lts.conf variable name then
20:13
<highvoltage>
vagrantc: ok. so the clearlooks gtk2 theme is still maintained upstream?
20:13
<alkisg>
LDM_SESSION_NAME vs LDM_SESSION, for example
20:13* vagrantc is weary of people changing the meaning of variable names
20:14
<alkisg>
Or LDM_SESSION vs LDM_SESSION_COMMAND, if we think that LDM_SESSION==the .dmrc session line
20:14
<vagrantc>
highvoltage: apparently no updates since october of 2010 there either
20:15
alkisg: that's not what it started out as..
20:15
it started out simply being a command to pass to Xsession
20:16
which is how it's implemented
20:16
which means the .dmrc should contain the reference to the .desktop file
20:17
and there's no technical conflict there, just that it's confusing to people.
20:17
like many things, it's a mess :)
20:19floorduh has left IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124, Ping timeout: 245 seconds)
20:19
<highvoltage>
vagrantc: I think I might have a solution soon that will satisfy both of us...
20:19
<vagrantc>
highvoltage: i like the idea of a smaller dependency chain that looks reasonably nice.
20:20gado has left IRC (gado!5f0f6c5d@gateway/web/freenode/ip.95.15.108.93, Ping timeout: 245 seconds)
20:21
<vagrantc>
highvoltage: i just installed gtk2-engines-cleanice and did sed -i -e 's,clearlooks,cleanice,g' greeter-gtkrc and it looks a little boxy, but is maybe working.
20:22
<alkisg>
vagrantc: ok, so I'll just try to fix dmrc handling so that it returns the whole command, and that it saves only the session name
20:24
<vagrantc>
i *think* that's the proper way to do it...
20:25
basically, .dmrc needs to be compatible with other display managers ...
20:25
what we do with that data may vary.
20:25
<alkisg>
Well, if we want something simpler for lts.conf, we can invent a new name, like LDM_SESSION_NAME
20:26
<vagrantc>
and then that looks for the corresponding .desktop file?
20:26
<alkisg>
I'll just fix it the way you said for now, it's a smaller fix :)
20:26
Yes
20:26
<highvoltage>
vagrantc: I was looking at how far we could get with a plain gtk + gtkrc, but it's hard to get around the boxyness with that
20:26
<vagrantc>
highvoltage: just go without a gtkrc at all
20:27
is it possible to detect which engines are available from within the gtkrc ?
20:27
<highvoltage>
vagrantc: we can't in ubuntu, otherwise we get the ugly resize handlebars
20:27
<vagrantc>
resize handlebars?
20:28
<highvoltage>
vagrantc: https://launchpadlibrarian.net/75959455/resize.png
20:29
(I guess it was a teacher who filed the bug, they like to ring things in red)
20:29
<vagrantc>
heh
20:32
highvoltage: dunno how to tweak gtkrc, but the cleanice one i generated by substituting cleanice for clearlooks looks a little clunky
20:32
don't know if that's the engine, a bunk gtkrc, or what.
20:34
<highvoltage>
vagrantc: ok, I guess for now I'll try to fix that issue while still using clearlooks
20:35
<vagrantc>
highvoltage: oh, that .png is an issue with clearlooks?
20:36artista_frustrad has left IRC (artista_frustrad!~fernando@200.247.43.2, Quit: Leaving)
20:36
<highvoltage>
vagrantc: it happens in other themes too, but in them I can work around it in the gtkrc files. with the clearlooks gtkrc file it didn't work. I'm taking some more time now to see why
20:37floorduh has joined IRC (floorduh!b85bfd7c@gateway/web/freenode/ip.184.91.253.124)
20:38
<vagrantc>
highvoltage: what about symlinking directly to the gtkrc shipped with clearlooks?
20:38
seems to work fine on debian...
20:39markit has joined IRC (markit!~marco@88-149-177-66.staticnet.ngi.it)
20:39
<highvoltage>
vagrantc: yeah the ugly handlebars thing is ubuntu-specific
20:40
(ubuntu adds them on all gtk windows by default, whish isn't nice for the invisible windows in ldm)
20:40* vagrantc wonders a bit why we ship a gtkrc
20:40
<vagrantc>
seems like the included ones work
20:41
<highvoltage>
I tried it with a very, very minimilistic one just now and it didn't actually look too bad
20:45
hmm, it works now. perhaps I just had a stale nbd before :/
20:46
<vagrantc>
i don't see a significant difference using the ldm greeter-gtkrc using clearlooks vs. the clearlooks gtkrc shipped with gtk2-engines
20:48
guess it's a little darker, and there's a little less boxiness aboutt the login box.
20:48
<markit>
btw, nbd image compressio is done with what algorithm? I've read wonderful things about LZ4
20:49
http://code.google.com/p/lz4/
20:49
<highvoltage>
heh, it didn't actually work. the new background I used just made it more subtle: http://irc.jonathancarter.org/files/screenshots/ldm-handles.png
20:50
<markit>
highvoltage: I entered too late, do you have the receipe to change login image easely?
20:51
<highvoltage>
markit: first, preheat your oven to 220°c
20:51
<markit>
ok, done
20:51
<highvoltage>
markit: the ldm themes are in /usr/share/ldm/themes in the ltsp chroot. the images are in the theme directories, you can change them there and then update your image :)
20:51
<markit>
(monitor is melting...)
20:51
<highvoltage>
(that means it's working)
20:52
<alkisg>
I'm rewriting X50-dmrc-processing, does anyone really use LDM_GLOBAL_DMRC, or should I remove support for it? I only see a code reference for it in K12Linux, but we have lts.conf options for LDM_SESSION and LDM_LANGUAGE, it shouldn't be needed... (and also it's broken now, so I don't think we're breaking anything :))
20:52
<markit>
highvoltage: so simple a matter to change a jpg? too easy, you are joking ;P
20:52
<vagrantc>
alkisg: i *think* that might have been fedora
20:52
<markit>
highvoltage: thanks for the tip
20:52
<highvoltage>
markit: nope, it's that easy.
20:53
<alkisg>
ltsp-trunk/server/configs/k12linux/lts.conf: LDM_GLOBAL_DMRC=/etc/ltsp/ldm-global-dmrc
20:54
...but if it's not working, and it's not needed, do we really need it in the code? Overhead without functionality..
20:54
<markit>
btw, with the new ltsp 5.3 will it be possible to have multiple LOCAL_APPS_EXTRAMOUNTS dirs?
20:54
<vagrantc>
alkisg: pfft.
20:54
<alkisg>
markit: that was possible in ltsp 5.2 too...
20:54
<markit>
alkisg: mmm we tried and did not worked, or I reacall badly?
20:55
<vagrantc>
alkisg: dunno. wish we had more activity from the fedora folks ... :(
20:55
<markit>
also would be nice to have the feature for NFS (I know only the option for NFS_HOME)
20:55
<alkisg>
vagrantc: I'll remove it and they can put it back if they become active :) It's not like it's working now anyway
20:55
<vagrantc>
does LOCAL_APPS_EXTRAMOUNTS rely on sshfs? it might be broken in debian now... sshfs doesn't seem to be able to mount dirs that you don't have full write access to anymore :(
20:55
<alkisg>
markit: LTSP 5.3 supports "FSTAB_0" to "FSTAB_9"
20:55
Complete lines to put in your fstab, so you can put nfs mounts there
20:56
<markit>
alkisg: so great
20:56
<alkisg>
markit: NFS_HOME also works but it's deprecated, so use FSTAB_x for that too
20:57
<markit>
hope someone will find the time to document these wonders of 5.3 (I can act as "tester" for the doc... if I can setup things reading the doc, then it's idiot-proof)
20:57
<vagrantc>
alkisg: what's with the break= code in init-ltsp.d ? isn't break used by initramfs-tools ?
20:57
<alkisg>
You can also have local /home mounted with FSTAB_x
20:57
<highvoltage>
vagrantc: I stripped more and more out of the grkrc file until those handles dissappeared. It worked, now I just need to do a diff and re-add things until their back :)
20:57
<alkisg>
vagrantc: it is, but only for break=premount etc, while ours have a digit in front, 50-x, and I thought it made sense to use the same namespace... break=xxx...
20:58
<vagrantc>
highvoltage: seems like it would be good to strip as much out of the default as we can ...
20:58
alkisg: feels like repurposing something and we might get burned by it.
21:00
<alkisg>
vagrantc: ok, let's make it ltsp.break=xxx
21:00
Like the kernel does with modules
21:01skumlesen has left IRC (skumlesen!~skumlesen@mail.kia.dk, Quit: crunchbang linux ftw)
21:02
<vagrantc>
alkisg: sounds reasonable ... although for some reason . seems like it'd be harder to match against, given that it's a regex...
21:02
alkisg: what about ltspbreak or ltsp_break ?
21:02
<highvoltage>
vagrantc: removing both these lines and adding my grip-{height,width} lines did the trick:
21:03
widget_class "*.tooltips.*.GtkToggleButton" style "clearlooks-tasklist"
21:03
widget "gtk-tooltips" style "clearlooks-tooltips"
21:03
not sure exactly how they import those settings again, but they do
21:05
<alkisg>
vagrantc: I don't like ltspbreak... ltsp_break sounds OK, I guess, but I still prefer the dot, like the kernel convension... no strong feelings, whatever you like. I don't think matching will be much of a problem.
21:06
<vagrantc>
alkisg: i didn't think it'd be a huge problem either, just came to mind initially.
21:06
highvoltage: can you commit to ldm-trunk and i can test on debian?
21:07
highvoltage: or if you'd rather, paste your gtkrc somewhere i can grab it
21:12komunista has left IRC (komunista!~slavko@adsl-195-098-015-242.dynamic.nextra.sk, Quit: Leaving.)
21:14
<highvoltage>
vagrantc: http://irc.jonathancarter.org/files/debian/paste/greeter-gtkrc
21:15
(that's probably way more gutted than necessary, not sure if it will result in ugliness that I missed)
21:15* vagrantc wonders if tftp can distinguish between no file and an empty file ... fetching lts.conf from tftp is not always desireable.
21:16
<highvoltage>
vagrantc: would you consider replacing the ldm background too? That blue background doesn't really work.
21:17
<vagrantc>
highvoltage: how badly does it not work? it's mostly supposed to be a very small minimalist example theme.
21:19
i mean, it seems to work fine as far as i can tell...
21:19
<highvoltage>
vagrantc: the contract between the text and the dark blue make it hard to read. the colourful logo doesn't look so great in front of the blue. I think a grey gradient like http://irc.jonathancarter.org/files/debian/paste/bg.png works better
21:19
the contrast is my biggest problem with it
21:19
you have to have good contrast wherever you have text.
21:20
with images it's still kind of forgivable sometimes. but not with text.
21:20
<vagrantc>
highvoltage: i'll check it out ...
21:21
highvoltage: your minimal greeter-gtkrc has the boxy problem...
21:21
highvoltage: no rounded edges ... at least on sid.
21:22
<highvoltage>
vagrantc: if you compare the background on http://irc.jonathancarter.org/files/screenshots/ldm-handles.png and https://launchpadlibrarian.net/75959455/resize.png then you can see how the text on the bottom has better contrast with the bg.png that I pasted
21:22
vagrantc: ok, I'll start again with a full one
21:24
<vagrantc>
highvoltage: your bg.png is 48k, whereas the original is 4k ...
21:25
will test what it actually looks like, though :)
21:25
<highvoltage>
vagrantc: I can get it down in size
21:25
<vagrantc>
highvoltage: though it does look nicer overall
21:29
although grey sure feels gray.
21:29
<highvoltage>
I guess it would work if I substitute the gray with blue
21:30
<vagrantc>
i don't know how the original image was made to be *so* small.
21:30
the original one i hacked up was about 100k
21:31
<highvoltage>
is the original one indexed? I guess it's also pngcrushed
21:31
<vagrantc>
no clue, digital artwork is not my forte by a long stretch
21:39skumlesen has joined IRC (skumlesen!~skumlesen@87-104-113-212-dynamic-customer.profibernet.dk)
21:51
<highvoltage>
in gimp I get it down to 40K and whatever I do in pngcrush it stays the same
21:51
(so obviously it's not my forté either :p)
21:51
<vagrantc>
heh
21:52
highvoltage: you've tried different colors yet?
21:54mischko has joined IRC (mischko!~schapman@205.247.25.89)
21:55
<highvoltage>
vagrantc: yep, switched to blue
21:55
brb, about to enter a flamewar
21:55
<stgraber>
highvoltage: you joined it pretty late, I've been fighting alone for a while...
21:56mistik1 has left IRC (mistik1!mistik1@unaffiliated/mistik1, Ping timeout: 244 seconds)
21:57mistik1 has joined IRC (mistik1!mistik1@unaffiliated/mistik1)
21:58
<vagrantc>
oho no
21:59
<highvoltage>
stgraber: sorry, I only noticed the conversation when I was highlighted
21:59
<stgraber>
hehe :)
22:01
<markit>
btw, the "preference" menu is not very intuitive, or at least would love to have a easy to find "turn off" button
22:02
at least, is the only function "my students" at school use when they log off from KDE
22:02
and are back to ldm
22:03mistik1 has left IRC (mistik1!mistik1@unaffiliated/mistik1, Ping timeout: 240 seconds)
22:03Trixboxer has left IRC (Trixboxer!~Trixboxer@115.124.115.71, Quit: "Achievement is not the end, its the beginning of new journey !!!")
22:03
<vagrantc>
tftp downloads an empty file if no file was found...
22:04mistik1 has joined IRC (mistik1!mistik1@unaffiliated/mistik1)
22:08
<highvoltage>
stgraber: and it also leaked over to a private discussion with jono on what "bullying" means. *sigh*
22:08
<stgraber>
highvoltage: I was pretty sure that'd happen ;)
22:10mistik1 has left IRC (mistik1!mistik1@unaffiliated/mistik1, Ping timeout: 255 seconds)
22:11mistik1 has joined IRC (mistik1!mistik1@unaffiliated/mistik1)
22:12markit has left IRC (markit!~marco@88-149-177-66.staticnet.ngi.it, )
22:19mistik1 has left IRC (mistik1!mistik1@unaffiliated/mistik1, Ping timeout: 260 seconds)
22:19mistik1 has joined IRC (mistik1!mistik1@unaffiliated/mistik1)
22:28
<highvoltage>
vagrantc: I guess this is less gray enough? http://irc.jonathancarter.org/files/debian/paste/ldm-blue.png
22:29
that tone of blue looks a little CGA though... I think I'll just change that a bit...
22:32
vagrantc: it's kind of weird. the default ltsp ldm theme looks boxy on ubuntu by default
22:34dead_inside has left IRC (dead_inside!~dead_insi@76.75.3.174, Quit: Leaving...)
22:35
<vagrantc>
highvoltage: do you have gtk2-engines installed? does it include gtk2-engines-clearlooks?
22:35
highvoltage: do you install without recommends?
22:35
<alkisg>
Yup, to that last one :)
22:35
<vagrantc>
since it's only recommended in debian, as it functions fine without it, and doesn't force people to install gtk2-engines
22:36* highvoltage installs it manually
22:36
<vagrantc>
i've still got a few annoying recommends...
22:36
but overall, installing without recommends seems like a really bad idea.
22:37
<highvoltage>
really? the first thing I do on a new debian machine is disable auto recommends installation
22:37
<vagrantc>
wow.
22:37
<alkisg>
Btw while testing logins, one time I didn't get a keyboard cursor at ldm and I couldn't type my username, then I dragged that edge handle that we want to get rid of, and then the keyboard cursor showed up
22:37
<vagrantc>
i guess i also selectively disable recommends when something's pulling in too much by default.
22:37
which is not something i'd expect your typical user to figure out
22:38
<highvoltage>
if I don't do that then I'm bound to have exim3 and mutt installed (and metric shitloads of other stuff) minutes after my installation is done
22:38* vagrantc is still fighting with http://bugs.debian.org/660297 :(
22:38
<highvoltage>
and just installing something like gdm3 pulls in the entire gnome if you have recommends enabled
22:38
(which is kind of sucky if you just want a good login thingy for xfce)
22:38
<vagrantc>
guess init-ltsp.d could mount them if not already mounted...
22:39* vagrantc doesn't use a display manager
22:39
<vagrantc>
except on other machines...
22:44
<highvoltage>
vagrantc: what do you think of this one? http://irc.jonathancarter.org/files/debian/paste/ldm-blue.png
22:45
(I'm sure since it's a gradient it should be possible to get it quite small again)
22:45risca has joined IRC (risca!~risca@wi-secure-8039.cc.umanitoba.ca)
22:46
<vagrantc>
highvoltage: will try it in a moment
22:46
highvoltage: rebuilding my LTSP chroot
22:47
highvoltage: oh, that's the whole image...
22:48
highvoltage: looks fine to me.
22:48
highvoltage: just a little lighter color, basically?
22:48
<highvoltage>
here's the background file: http://irc.jonathancarter.org/files/debian/paste/bg-blue.png
22:48
(still a bit big at 57k)
22:49
lighter than the original, darker than the one I pasted previously
22:49brunolambert has left IRC (brunolambert!bruno@nat/revolutionlinux/x-mybpgqekwjqxticc, Quit: brunolambert)
22:49
<highvoltage>
the ligher blue looked like a CGA blue, which I usually try to avoid
22:50
<vagrantc>
still no luck squishing it further? dunno what the artist of that tiny file did to make it so small.
22:51
debian's ltsp will finally have a fat-client plugin ...
22:51
though i didn't bother with any "sane" default, i rely on the admin to select one.
22:51
<highvoltage>
I can try... pngcrush just gave me the same filesize back. I guess there's an option somewhere to optimize for a verticle gradient... I'll have to dig a bit
22:51
ah great
22:52
<alkisg>
$ file /opt/ltsp/i386/usr/share/ldm/themes/ltsp/bg.png
22:52
/opt/ltsp/i386/usr/share/ldm/themes/ltsp/bg.png: PNG image data, 640 x 480, 8-bit colormap, non-interlaced
22:52
<vagrantc>
would probably be smart to check for common ones like gnome, kde, xfce and lxde...
22:52
but it's so hard to track which specific package should pull those in.
22:53
<highvoltage>
alkisg: mine is 8 bit too
22:55
<vagrantc>
the debian kernel team said they'd drop support for /boot/vmlinuz and /boot/initrd.img symlinks ... but they haven't yet ... ugh. tracking specific kernel versions is a pain i don't want to have to implement.
22:55
the easiest workaround, i guess, would be to implement the symlinks in LTSP ...
22:56
although i have use-cases for amd64 vs. pae vs. i386 now at freegeek, so supporting multiple selectible/detectible kernels would be ideal.
22:57
ifcpu64.m32 should allow handling that...
22:57* vagrantc is tempted
22:58
<vagrantc>
alkisg: we need to re-implement the pxelinux configuration code pretty much from scratch.
22:58
what a mess.
22:59
<highvoltage>
ok so the optimization I thought exist does and it's called "scanline filtering" http://www.smashingmagazine.com/2009/07/15/clever-png-optimization-techniques/
22:59
now I just need a tool that can do it :)
22:59
<alkisg>
highvoltage: how you create the file also matters
22:59
<vagrantc>
alkisg: the direction i was going when LTSP stole my attention away from lessdisks was to have some sort of plugin system, so that it was easy to override if the defaults were wrong, or the specific sub-architecture needed something crazy.
22:59
<alkisg>
E.g. if you start with the original one, convert to RBG, change contrast, and save, (with GIMP), you still get < 2 Kb
23:00
<vagrantc>
highvoltage: when i searched for pngcrush, there appeared to be several other packages that came up that might do similar or better things.
23:00
<highvoltage>
alkisg: < 2 kb!?
23:00* highvoltage tried optipng
23:00
<alkisg>
vagrantc: I think we need to talk with the pxelinux packagers there, and implement something they'll be willing to accept
23:01
<vagrantc>
alkisg: that would be ideal... alternately, just implement our own hooks that use pxelinux.
23:01
<alkisg>
highvoltage: e.g.: http://imagebin.org/index.php?mode=image&id=199924
23:01
<vagrantc>
alkisg: in their own package...
23:01
<alkisg>
1756 bytes
23:01
<vagrantc>
i think someone sold their soul to create such a small .png.
23:02_UsUrPeR_ has joined IRC (_UsUrPeR_!~jsass@c-76-112-192-21.hsd1.mi.comcast.net)
23:03
<_UsUrPeR_>
hello all
23:03
<highvoltage>
what the heck... why is mine so big
23:04
<alkisg>
highvoltage: ah, what are you using? Gimp?
23:04
In the gradient tool there's an option for dithering
23:04
Uncheck it :)
23:04
<stgraber>
highvoltage: I just had a pretty constructive conversation with Michael Hall, it's really easy to discuss ARB problems when Jono isn't around ;)
23:04
<alkisg>
2 kb without it :)
23:04
<stgraber>
highvoltage: anyway, sent a couner-proposal to the ARB ML
23:05
<alkisg>
Hi _UsUrPeR_
23:05* _UsUrPeR_ waves
23:07xsl has joined IRC (xsl!~silence@unaffiliated/xsl)
23:09
<highvoltage>
alkisg: yeah, gimp
23:09
<alkisg>
highvoltage: by unchecking dithering, you get very small file sizes
23:10
<highvoltage>
alkisg: yay! alkisg saves the day again
23:11
<vagrantc>
and it still looks nice enough?
23:11
<alkisg>
highvoltage: also, I think you should grow it to 1024x768 and use real RBG, not indexed color
23:11
Even if it makes it 2K
23:12
Because otherwise the horizontal lines look bad
23:12
*if it makes it 4K instead of 2, sorry
23:12* vagrantc 's first go at a fatclient test install failed, hopefully due to transient problems in sid rather than any fundamental problems
23:12
<highvoltage>
I tend to agree with you there. I think vagrantc will be ok with 2K especially if it looks a lot better
23:12
<vagrantc>
4k vs 2k, no big difference.
23:13
40k, that's not so ideal.
23:13
<alkisg>
We could also write code in ldm to draw the gradient itself :P
23:13
Faster, saves ram too
23:14
<vagrantc>
considering most distros will probably default to their own custom theme, no?
23:14
that's why we want it so small.
23:14
<highvoltage>
alkisg: yeah then you could set the gradient colours as lts.conf options too :)
23:14
bg.png 4434
23:14
is that ok?
23:14
<vagrantc>
alkisg: that would really be faster?
23:14
<alkisg>
vagrantc: based on my experience with windows api, yes, lots
23:14
<vagrantc>
highvoltage: fine by me!
23:15
alkisg: that's not just bad image handling?
23:15
alkisg: how complicated are we talking, that sounds like an awesome way to do it if it's really faster, smaller disk size, and more flexible.
23:16
<highvoltage>
vagrantc: http://irc.jonathancarter.org/files/debian/paste/bg-blue.png
23:16
<alkisg>
I've no idea about gtk or X calls, but it should be less than 10 lines to draw a gradient background
23:17
highvoltage: much better than the old one :)
23:18
<highvoltage>
yay
23:19
<alkisg>
What would be really fun, is to have an animated screensaver-like wallpaper for ldm background :D
23:19* alkisg did some of these back in the dos/assembly days...
23:19
<vagrantc>
alkisg: now you're bringing your windows background a bit too much into this
23:19
<highvoltage>
heh, like the 4k demos the crackers used to do
23:19
<vagrantc>
highvoltage: background looks good to me! commit it!
23:19
<alkisg>
Yup, one of those
23:20
<highvoltage>
vagrantc: you have to commit it. I don't have ltsp upstream rigtts :)
23:20
<alkisg>
We need to fix that
23:20
<vagrantc>
highvoltage: that's ridiculous!
23:20
<alkisg>
Anyone who has had lobster with us should have commit rights :P
23:20
<highvoltage>
heh, nice requirement :)
23:21
ooh, except that I don't actually eat lobster :)
23:21
I guess that explains a few things
23:21
<alkisg>
-1 :D :D :D
23:21
Meh, ok, it was a pitty seeing them alive first and roasted 20 minutes later
23:22
<vagrantc>
highvoltage: added.
23:22
alkisg: a pity? it's the proper way to eat things.
23:22
highvoltage: go commit it yourself! :P
23:23
<highvoltage>
vagrantc: whohoo! I'm going to have to celebrate :)
23:23
(but first I need to go home :p)
23:23
<vagrantc>
i guess technically stgraber should've added highvoltage, but highvoltage has been working on debian package stuff too...
23:23
<highvoltage>
stgraber had plenty of opportunities to add me :)
23:23
<stgraber>
vagrantc: I can't add him, I'm not an admin of the team ;)
23:23
<highvoltage>
ah, I stand corrected
23:24
<vagrantc>
hah!
23:24
https://launchpad.net/~ltsp-upstream/+mugshots
23:25
stgraber: that seems wrong, too.
23:25
stgraber: i'm guessing it's still ogra?
23:26
stgraber: well, i hope you're happy i added him :)
23:26
<stgraber>
vagrantc: yep, I'm definitely fine with that ;)
23:27
<vagrantc>
highvoltage: so should we wait till you get a chance to commit it yourself? or should i just go ahead and push
23:28
<stgraber>
vagrantc: yeah, I need to bribe ogra to fix that ;) usually beer works fine for that
23:29
<vagrantc>
stgraber: heh
23:30
hrm. fat client support isn't working.
23:31
oh, it's that pesky initramfs-tools bug again, probably.
23:31
time to implement a workaround
23:41loather has left IRC (loather!~khudson@wsip-98-175-250-115.sd.sd.cox.net, Quit: This computer has gone to sleep)
23:43xsl has left IRC (xsl!~silence@unaffiliated/xsl, Remote host closed the connection)
23:44
<alkisg>
ssh -S /var/run/ldm_socket_8034_192.168.0.1 192.168.0.1 grep -lR "^Exec=gnome-session\\ --session=ubuntu-2d" /usr/share/xsessions/
23:44
Anyone knows of a way to avoid that \\ there?
23:46
<vagrantc>
alkisg: do you want an exact match anyways?
23:47
<alkisg>
vagrantc: the problem is possibly that ssh uses "$*" instead of "$@"
23:47
So I can't pass "LDM_SESSION" to grep properly
23:48
So if I can't find a way, I'll have to slash-escape the LDM_SESSION before passing it to ssh...
23:51
<vagrantc>
alkisg: could do fuzzy matches and parse locally
23:52
<alkisg>
True... let me try that
23:53
<vagrantc>
are there xsessions buried more than one layer deep?
23:58
<highvoltage>
vagrantc: heh, that looks just like you!
23:58
vagrantc: I'll update the background :)
23:59
<alkisg>
vagrantc: no, the -R is there so that we don't pass * to ls