2009-12-01

For minutes of previous meetings, please see DesktopTeam/Meeting.

Meeting Minutes

Present

Main Meeting

  • Rick Spencer (rickspencer3) - chair

  • Arne Goetje (ArneGoetje)

  • Bryce Harrington (bryce)
  • Chris Cheney (ccheney)

  • Ken VanDine (kenvandine)

  • Martin Pitt (pitti)
  • Sebastien Bacher (seb128)
  • Till Kamppeter (tkamppeter)
  • Alberto Milone (tseliot)

Eastern Edition

  • Robert Ancell (robert_ancell)

Apologies

  • Jonathan Riddell (Riddell) - holiday

  • Luke Yelavich (TheMuso) - moving house

Agenda

  • some manager announcements
    • Welcome tselliot
    • 1-1 calls
    • UNE and didrocks
    • receipts and robert_ancell's scanning app
  • Blueprints, work items, milestones, specs (pitti)
  • Ideas for speeding up desktop startup time (seb128)

Actions from this meeting

  • ACTION: rickspencer3 to track down robert's scanning tool
  • ACTION: seb128 to add initial work items to UNE session blueprint
  • ACTION: robert_ancell to do seed changes for new games
  • ACTION: rickspencer3 to confirm robert_ancell commitment to GDM/GDM Greeter refactor
    • RESULT: Based on current OEM commitment, documenting the current API is in scope for A2, rest of the work may be done post A2
  • ACTION: kenvandine to investigate improving startup time of Messaging Menu and related tasks
  • ACTION: ricspencer3 to set up call with pitti, bryce, and tselliot to get tselliot on board
  • ACTION: all to try robert_ancell's new scanning tool

some manager announcements

  • Welcome back tselliot here on OEM team rotation
    • bryce will be his "buddy"
    • will work on xorg drivers and packaging gnome initially
  • weekly 1-1 calls: please set these up with me if they are not on your calendar
  • UNE and didrocks: UNE will be moved to desktop team, and didrocks will be the maintainer
  • receipts and robert_ancell's scanning app: expenses coming up, please give robert's app a try and file bugs

Blueprints, work items, milestones, specs (pitti)

pitti covered A2 targeted blueprints that had no work items, or otherwise seems risky:

rickspencer3 closed this part of the meeting by stressing that it is better to pull work items in from A3 than to postpone them from A2, so encourages team members to be reasonably conservative in their work item targeting for A2.

Ideas for speeding up desktop startup time (seb128)

seb128 led a session on improving load time for the desktop.

  • budget for desktop load time is 4 seconds from GDM

  • a current boot chart is available here: http://people.canonical.com/~seb128/bootchart/seb128-dellmini-lucid-20091201-1.png

  • with nautilus and xsession.d both not loaded at boot time, time is still 8 seconds, so need to get back 4 seconds
  • pitti is working on caching application data for application menu, should buy some time
  • other ideas for saving time were discussed, some at length. Example: removing the desktop background image saves .5 seconds. xsplash takes .8 seconds
  • Current plan is to remove many applets, and to load with only one panel, and to use metacity rather than compiz by default

Activity reports

Alberto Milone (tseliot)

  • foundations-lucid-boot-experience spec:
    • Write a Plymouth plugin which can replace Usplash features - In progress
  • desktop-lucid-xorg-proprietary-drivers spec:
    • Start looking at the debian packages for nvidia for a first clean up - TODO
  • desktop-lucid-touchscreen-handling spec
    • Identify one or more target touchscreen hardware platforms to focus on - DONE
  • becoming core-dev - In progress:
    • recommendation from the MOTU council: DONE
    • approval from the Technical board: TODO

Arne Goetje (ArneGoetje)

  • Blueprints:
    • desktop-lucid-font-selection: pending approval
    • desktop-lucid-ibus-chinese: need more research
    • desktop-lucid-language-selector: need input from design team
  • research about XKB for the desktop-lucid-ibus-chinese spec
  • upload latest Firefox localisation XPIs into Rosetta for all supported releases
  • filing bugs about buggy Firefox XPIs upstream

Bryce Harrington (bryce)

  • Off Thanksgiving Holiday - Nov 26, Nov 27
  • Completed drafting and work item identification on blueprints:
    • desktop-lucid-xorg-triaging-diagnosis
    • desktop-lucid-touchscreen-handling
    • desktop-lucid-xorg-driver-selection-for-nvidia-hardware
    • desktop-lucid-xorg-proprietary-drivers
  • Investigate resolution detection issues.
    • Seems with -intel KMS some quirks were not moved to the kernel,
      • and so EDID fails in some cases.
    • Wrote up
  • Begun drafting a how-to fix X.org crashes guide
  • X.org planning for Lucid
  • Updated versions_current.html for Lucid merges, add Debian Testing

Chris Cheney (ccheney)

  • Off vacation
  • Off Thanksgiving Holiday
  • OOo bug triage

Ken VanDine (kenvandine)

  • Finished specs
  • Cleaned up gwibber package deps, no longer depend on python-gnome2-desktop, just python-wnck
  • Did some work with the desktopcouch branch of gwibber
  • Triaged gwibber bugs and merge proposals (before duplicating work in the spec)
  • DX and U1 planning/review

Martin Pitt (pitti)

Drafted blueprints:

  • desktop-lucid-default-apps, desktop-lucid-jockey-hotplug-support, desktop-lucid-suspend-quirks-halsectomy, desktop-lucid-xorg-halsectomy: approved
  • desktop-lucid-likewise-open-migration, desktop-lucid-bug-management: Ready for review by Rick

Assigned blueprints:

  • desktop-lucid-likewise-open-migration: blocked on upstream sending the packages
  • desktop-lucid-jockey-hotplug-support (beta available): Should be working, just a bug fix and documentation left
  • desktop-lucid-suspend-quirks-halsectomy: Discussed migration with upstream, git branch available
  • desktop-lucid-xorg-halsectomy: I committed the required udev helper upstream; created/discussed udev rules with Julien Cristau, Debian experimental/Lucid PPA available
  • desktop-lucid-powermanagement-tweaks: No progress this week, needs drafting first

Non-blueprint work done:

  • Improved, documented and announced work item handling
  • fixed gdm to get along without hal
  • fixed remaining hal-isms in pm-utils (other than suspend quirks)
  • Bug fixing: apport, udev
  • Went through our pm-utils delta, committed most of it to Debian, moved remaining things to the right place, forwarded stuff upstream; we are back in sync now
  • Reviewed desktop team specs

Merges: None done this week, but I did most of them already

Sponsoring:

  • pidgin (intrepid SRU), pkg-create-dbgsym

Robert Ancell (robert_ancell)

  • Finish off specifications
  • Lucid packaging gcalctool, eog, gucharmap, brasero, evince, gnome-games, rhythmbox, xscreensaver, metacity, gnome-utils, gnome-terminal

Sebastien Bacher (seb128)

blueprints:

  • desktop-lucid-startup-speed: drafted, approved, started work
  • desktop-lucid-gnome-3: informational spec about GNOME changes for lucid, drafted, approved
  • desktop-lucid-compiz-effects: drafted, approved, no work on it yet
  • desktop-lucid-gnome-geoclue: drafted, approved, deferred to next cycle the stack is not ready

non blueprint work:

  • desktop merges and GNOME 2.29 updates: gnome-pkg-tools gconf gtksourceview gedit tomboy poppler gftp gimp
  • telepathy-gabble karmic sru to fix some proxy server overuse
  • did lucid syncs and autosyncs after uds, lot of those and lot of issues due to the new debian source format, took quite some time
  • sponsoring: tomboy gnome-session evolution-indicator(sru)

login speed:

  • set up my dellmini as a testing box and did a lot of bootcharting
  • started the cycle with a 25.3 seconds boot time, 15 seconds desktop time
  • weekly boot time: 22.6 seconds, 12.9 seconds desktop login
  • desktop login time without compiz, nautilus nor xsession.d script: 8 seconds
  • lot of delay are due to gnome-session waiting for registrations before starting other things

  • compiz and pulseaudio wrappers have been dropped (thanks TheMuso Amaranth): 1 second win

  • compiz-fusion-plugins-extra doesn't need to be installed by default: will win 0.5 second
  • chrisccoulson did work on gnome-session and gnome-settings-daemon, nothing uploaded yet but some small wins already and ideas on other possible changes
  • tried to drop gnome-session delayed starts, the boot chart looks compact but there is almost no change to the login time, too many concurrential runs that's clearly not optimal

Till Kamppeter (tkamppeter)

  • Uploaded new HPLIP to merge from Debian and to fix LP: #335662. An FTBFS occured. Investigated it.
  • Investigated also a CUPS FTBFS finding out that in Poppler the API and ABI was changed in the middle of a stable series and without bumping the soname. Got patch to fix the problem from Koji Otani (upstream author of libpoppler-based "pdfto..." filters).
  • Answered and triaged printing-related bug reports.

IRC Log

[16:30] <rickspencer3> ArneGoetje, bryce ccheney kenvandine pitti seb128 Riddell tkamppeter tseliot
[16:30] <rickspencer3> https://wiki.ubuntu.com/DesktopTeam/Meeting/2009-12-01<<BR>> [16:31] <rickspencer3> most of the meeting will be for pitti and seb128 but I have some managery announcements to get out of the way
[16:31] * rickspencer3 copies and pastes from wiki
[16:31] <rickspencer3> Welcome back tselliot here on OEM team rotation
[16:31] <rickspencer3> * bryce will be his "buddy"
[16:31] <rickspencer3> * will work on xorg drivers and packaging gnome initially
[16:31] * rickspencer3 hiya tseliot it's great to have you back
[16:32] * pitti hugs tseliot
[16:32] <seb128> tseliot, welcome!
[16:32] <tseliot> it's great to work with all of you again
[16:32] <kenvandine> :)
[16:32] <rickspencer3> we should have a call with pitti and bryce to discuss your work for the desktop team in Lucid
[16:32] <pitti> my nvidia card is fail plz halp
[16:32] <kenvandine> hehe
[16:32] <tseliot> hehe ok
[16:32] <rickspencer3> I remember meeting tseliot at my first UDS
[16:32] <rickspencer3> he had the best sessions, and he didn't even work at Canonical then
[16:33] <pitti> too bad that you couldn't come this time
[16:34] <rickspencer3> so the OEM rotation means that robert_ancell will be with the OEM team
[16:34] <rickspencer3> for Lucid
[16:34] <tseliot> pitti: yes, I'll try to make up for it by attending the sprint
[16:34] <rickspencer3> :)
[16:34] <pitti> tseliot: you have to cook pasta
[16:34] <rickspencer3> hehe
[16:34] <kenvandine> yum
[16:35] <tseliot> pitti: hehe, actually I'm better at eating pasta
[16:35] <rickspencer3> okay, we'lll set up a call
[16:35] <dobey> hehe
[16:35] <rickspencer3> which brings up ...
[16:35] <tseliot> good
[16:35] <rickspencer3> 1-1 calls: please set these up with me if they are not on your calendar
[16:35] <dobey> yeah, i was wondering where tseliot was in Dallas
[16:36] <tseliot> ;)
[16:36] <rickspencer3> I noticed that a lot of the 1-1 calls have fallen off my calendar, so tseliot and anyone who doesn't seem to have one, please pick a time that is good for you and schedule that
[16:36] <rickspencer3> ok, next
[16:37] <rickspencer3> # UNE and didrocks: UNE will be moved to desktop team, and didrocks will be the maintainer
[16:37] <rickspencer3> not sure everyone is aware of this change, so thought I should bring it up here
[16:37] <bryce> welcome to the team tseliot!
[16:37] <tseliot> rickspencer3: ok, I'll send you an email
[16:37] <tseliot> bryce: thanks
[16:37] * ccheney is here
[16:37] <seb128> didrocks, congrats again ;-)
[16:37] * ccheney didn't realize the meeting changed times to 16:20
[16:37] <rickspencer3> UNE is Ubuntu Netbook Edition (was Ubuntu Netbook Remix)
[16:37] <seb128> ccheney, it didn't
[16:38] <seb128> it started at :30
[16:38] <tseliot> didrocks: congrats
[16:38] <ccheney> hmm my system clock is broken then :(
[16:38] <rickspencer3> ccheney is in a space ship traveling at a very high speed
[16:38] <ccheney> wow my system is a full 10m behind
[16:38] <tseliot> :-D
[16:39] <rickspencer3> okay, so anyway, didrocks will be joining the desktop team focused on UNE
[16:39] <rickspencer3> we won't refuse any gnome packaging that he may do to help seb128, though
[16:39] <rickspencer3> okay, almost done with the boring manager stuff
[16:39] <rickspencer3> # receipts and robert_ancell's scanning app: expenses coming up, please give robert's app a try and file bugs
[16:39] <seb128> hehehe :-)
[16:39] <rickspencer3> robert has put together a sweet and simple scanning tool
[16:39] <seb128> doh, I already did those
[16:40] <rickspencer3> with expenses coming up, it seems a good chance to try it out
[16:40] <rickspencer3> tell your friends
[16:40] <tseliot> where can I find this tool?
[16:40] <rickspencer3> help robert out!
[16:40] <seb128> I will find something else to try it anyway ;-)
[16:40] <seb128> rickspencer3, we do we find it?
[16:40] <rickspencer3> tseliot, good question
[16:40] <seb128> we->where
[16:40] <rickspencer3> ACTION: rickspencer3 to track down robert's scanning tool
[16:40] <rickspencer3> I assumed it
[16:40] <pitti> I'll try it, I archive all my snail mail with it
[16:40] <and471> jono, hi
[16:40] <rickspencer3> s in his ppa, but not 100% sure
[16:40] <kenvandine> lp:simple-scan
[16:41] <and471> jono, I see on the ubuntu wiki you have added ##BEGINLERNID
[16:41] <kenvandine> hopefully in a ppa too
[16:41] <seb128> kenvandine, we want ppa
[16:41] <seb128> ;-)
[16:41] <pitti> https://launchpad.net/~robert-ancell/+archive/ppa<<BR>> [16:41] <rickspencer3> and471, hiya, we're having our team meeting atm
[16:41] <and471> jono, however I believe we can do it a simpler way
[16:41] <pitti> it's there
[16:41] <and471> rickspencer3, sorry
[16:41] <rickspencer3> and471, feel free to hang out, we'll be done soon
[16:41] <rickspencer3> and471, no apologies necessary, happens all the time
[16:41] <rickspencer3> ok, that's the boring managery announcements
[16:41] <rickspencer3> assuming there are no questions
[16:41] * rickspencer3 hands mic to pitti
[16:42] <pitti> thanks
[16:42] <pitti> so, let's look at the alpha-2 BPs
[16:42] <pitti> WARNING: desktop-lucid-unr-session has no work items
[16:42] <pitti> didrocks: I know that you are not yet on duty
[16:42] <pitti> do you want to set them up yourself, or want someone to help with that?
[16:43] <rickspencer3> pitti, I can get that one statred
[16:43] <rickspencer3> also I was thinking we would not do anything there in a2, since didrocks won't be started yet
[16:43] <rickspencer3> is that okay?
[16:43] <pitti> WFM
[16:43] <seb128> I can as well, I was initially drafter
[16:43] <pitti> it's not critical for alpha-2 anyway, it's not a very intrusive change wrt. to other parts of the distro
[16:43] <seb128> I just handed it to didrocks since he started on the wikipage before me
[16:43] <rickspencer3> ACTION: seb128 to add initial work items to UNR session spec
[16:44] <pitti> moved to beta-1 now
[16:44] <rickspencer3> thanks pitti
[16:44] <pitti> WARNING: desktop-lucid-powermanagement-tweaks has no work items
[16:44] <pitti> that's half my fault
[16:44] <pitti> I pinged Michael Frey again, he'll do the initial drafting now
[16:44] <rickspencer3> who's is the other half? (expect it is me)
[16:44] <pitti> and then I'll review and add WIs
[16:44] <rickspencer3> pitti, again, let's move that to a3?
[16:44] <pitti> rickspencer3: just that you are too gentle with your whip :)
[16:45] <pitti> rickspencer3: I'd like that to be in a2, since it does have intrusive changes
[16:45] <pitti> I just need to follow up on that
[16:45] <pitti> so this was by and large just a FYI
[16:45] <pitti> the others all have WIs
[16:45] <pitti> let's do a quick run-through and check for problems
[16:45] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-xorg-halsectomy<<BR>> [16:45] <pitti> this is well underway
[16:45] <pitti> bryce: what's your plan wrt. merging 1.7.1 from experimental?
[16:46] <pitti> since dropping hal from the default install will likely cause some regressions, it would be great to see that land soon
[16:46] <pitti> I already fixed g-p-m, but I'm sure that there are less visible ones
[16:46] <pitti> I'll talk to Keybuk about updating udev to latest upstream trunk, so that we have the building blocks
[16:47] <pitti> but, I have run my desktop with hal purged since Friday, works well \o/
[16:47] <pitti> (should buy us some ~ 1.5 seconds startup speed)
[16:47] <tseliot> \o/
[16:47] <bryce> pitti, I need to check with timo but as soon as debian feels it is good to go we should pull it in
[16:47] <seb128> 1.5 seconds for early boot and xorg right?
[16:47] <pitti> seb128: *nod*
[16:48] <seb128> no impact on desktop there?
[16:48] <bryce> I need to find out if timo intends to do that or if I should. But I hope to have it in this week
[16:48] <seb128> still good news ;-)
[16:48] <pitti> bryce: that would be great
[16:48] <pitti> otherwise this seems well underway and realisticc
[16:48] <bryce> pitti, was that "reduced to 1.5 sec" or "reduced by 1.5 sec"?
[16:49] <pitti> (modulo insurmountable regressions with wacom, of cousre)
[16:49] <pitti> bryce: "by"
[16:49] <bryce> pitti, from what approximately?
[16:49] <pitti> it's not _that_ bad :)
[16:49] <pitti> bryce: I'm not sure
[16:49] <pitti> 10ish
[16:49] <bryce> on my laptop (with SSD) it takes ~2 sec
[16:49] <pitti> (that's grub->gdm)
[16:49] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-default-apps<<BR>> [16:50] <pitti> this seems simple
[16:50] <pitti> would be great if the seeds could be changed soon for gnome-games
[16:50] <pitti> otherwise we need someone to do a MIR for "seed" (two new games pull it in)
[16:50] <pitti> rickspencer3: ^ would be a task for robert?
[16:50] <rickspencer3> pitti, would it be reasonable to ask robert_ancell to do that
[16:50] <rickspencer3> ?
[16:50] <rickspencer3> lol
[16:50] <rickspencer3> okay, gmta
[16:50] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-social-from-the-start<<BR>> [16:50] <rickspencer3> ACTION: robert_ancell to do mri for new games
[16:50] <pitti> this is a killer
[16:51] * rickspencer3 looks meaningfully at kenvandine
[16:51] <pitti> kenvandine: ^ this is currently targetted for a2, but it sounds like it'll take two manmonths to implement that all from reading the WIs
[16:51] <pitti> how bad is it really?
[16:51] <pitti> should we spread it out a bit to move some bits to beta1?
[16:51] <kenvandine> not that bad
[16:51] <pitti> rickspencer3: s/mri/seed changes/
[16:51] <kenvandine> i might end up defferring some to b1
[16:52] <kenvandine> but with the couch stuff, it actually greatly simplifies quite a bit
[16:52] <kenvandine> and much of that is done, but still experimental
[16:53] <pitti> ok; let's keep it on the a2 radar for now, but please speak up early if it's getting late
[16:53] <pitti> oh, good!
[16:53] <kenvandine> will do
[16:53] <rickspencer3> can we just move the least essential tasks to a3?
[16:53] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-xorg-proprietary-drivers<<BR>> [16:53] <kenvandine> rickspencer3, i kind of did
[16:53] <kenvandine> :)
[16:53] <pitti> tseliot: are there any blockers for this?
[16:53] <pitti> bryce, tseliot: and do we actually have a working nvidia driver for xorg 1.7/2.6.32, so that this can go forward and be tested?
[16:54] <tseliot> pitti: I'm waiting for superm1 to review a patch of mine for DKMS
[16:54] <tseliot> pitti: furthermore I've been busy with plymouth but I should be able to work on that soon
[16:54] <bryce> pitti, I believe the currently released one works with 1.7
[16:54] <pitti> nice
[16:54] * tseliot nods
[16:55] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-gdm-custom-greeter-support<<BR>> [16:55] <didrocks> pitti: seb128 : rickspencer3 : o/ and sorry for disturbing (just back home) I'm adding WI to the unr spec this evening (was thinking that the spec has to be approved first)
[16:55] <pitti> rickspencer3: ^ can you handle that in the eastern edition?
[16:55] <pitti> didrocks: merci! (no, it won't get approved before it has WIs Smile :) )
[16:56] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-likewise-open-migration<<BR>> [16:56] <rickspencer3> pitti, could you please add comments to the white board wrt any issues that you see with the spec?
[16:56] <pitti> this is basically blocked on getting new packages from upstream; nothing to discuss from my side, unless there are questions?
[16:56] <pitti> rickspencer3: no technical ones, just "does robert have time to do it?"
[16:57] <rickspencer3> ah
[16:57] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-xorg-triaging-diagnosis<<BR>> [16:57] <pitti> bryce: any blockers for that?
[16:57] <pitti> and do you think it's a realistic target for a2?
[16:57] <pitti> (42 work items..)
[16:57] <rickspencer3> ACTION: rickspencer3 to confirm robert_ancell commitment to GDM/GDM Greeter refactor
[16:57] <bryce> pitti, manpower mainly
[16:58] <bryce> pitti, well at least 50% of that can (and should) be done by a2
[16:59] <pitti> it seems that it'd be half as useful to do 50% of them, so it's still good to start it early
[16:59] <pitti> great
[16:59] <pitti> so, I think we are well on track
[16:59] <pitti> https://blueprints.launchpad.net/ubuntu/+spec/desktop-lucid-empathy-indicator<<BR>> [16:59] <kenvandine> yeah
[16:59] <rickspencer3> bryce, can we move the other 50% out of the a2 work item list?
[17:00] <pitti> kenvandine: so, from what I understand, this will be completely new code, and a new daemon, right?
[17:00] <kenvandine> pitti, did you read my response?
[17:00] <kenvandine> yes
[17:00] <pitti> kenvandine: I did
[17:00] <pitti> kenvandine: (no python please Smile :) )
[17:00] <bryce> rickspencer3, how?
[17:00] <kenvandine> easiest way to maintain it imho
[17:00] <kenvandine> vala!
[17:00] <pitti> kenvandine: FYI, I have example vala d-bus code, if you need it
[17:01] <kenvandine> pitti, do share
[17:01] <kenvandine> docs are weak :)
[17:01] <pitti> kenvandine: do you think you can squeeze it into your already full schedule, or should we postpone some social-from-start issues?
[17:01] <kenvandine> i will move some social stuff
[17:01] <pitti> kenvandine: http://www.piware.de/2009/11/hello-dbus-in-vala/ (I have some more, too)
[17:02] <kenvandine> thx
[17:02] <pitti> any other questions/
[17:02] <pitti> ?
[17:02] <rickspencer3> not a question, but a point I would like to make
[17:03] <rickspencer3> I would *much* rather pull things from a3 into a2
[17:03] <rickspencer3> rather than postpone things from a2
[17:03] <rickspencer3> so I expect everyone to do a reasonable job scoping their a2 work, and I consider "less is more" wrt a2 commitments
[17:03] <pitti> sure, it doesn't mean that you mustn't work on post-alpha2 work items :)
[17:03] <rickspencer3> exactly
[17:04] <rickspencer3> so my feedback is that it's better to be conservative in your estimates about what you can deliver in a2
[17:04] <rickspencer3> </soapbox>
[17:04] <pitti> mic back to rickspencer3
[17:05] <rickspencer3> pitti, thanks for doing a incredible job getting us organized before, during, and after UDS
[17:05] <rickspencer3> and the burndown charts are amazing
[17:05] * seb128 hugs pitti
[17:05] * pitti bows
[17:05] <bryce> rickspencer3, ok done
[17:05] <tseliot> burndown charts?
[17:05] <pitti> I'll adjust the trend lines on Thursday, when we are at the BP definition deadline
[17:05] <rickspencer3> ah, poor tseliot, a burndown chart newbie
[17:05] <pitti> tseliot: http://piware.de/workitems/desktop/lucid-alpha2/report.html<<BR>> [17:06] <pitti> tseliot: http://piware.de/workitems/desktop/lucid/report.html<<BR>> [17:06] <pitti> tseliot: check out the previous one (http://piware.de/workitems/desktop/karmic/report.html) for the full idea
[17:06] <tseliot> ok, thanks
[17:06] <rickspencer3> tseliot, we can talk on the phone, it's basically the platform team way to track status of work done compared to commitments
[17:06] <rickspencer3> thanks again pitti
[17:06] <pitti> tseliot: also, https://wiki.ubuntu.com/WorkItemsHowto has some docs
[17:06] <tseliot> ok
[17:07] * rickspencer3 hands mic to seb128
[17:07] <seb128> alright
[17:07] <seb128> so as everybody probably knows by now we have pretty agressive requirements about boot speed in lucid
[17:07] <seb128> I've started look at desktop login this week
[17:08] <seb128> our budget for desktop is around 4 seconds
[17:08] <seb128> current chart is http://people.canonical.com/~seb128/bootchart/seb128-dellmini-lucid-20091201-1.png<<BR>> [17:08] <seb128> (I will copy regular charts from the reference box on this webspace for those interested in that)
[17:09] <seb128> right now we are around 3 times slower than we ough to be
[17:09] <seb128> ie there is quite some hard work to put there and we might need to tweak our default config and selection
[17:09] <seb128> I would like to discuss changing the list of default applet for this cycle
[17:10] <pitti> so unless it's not too much work for speeding up compiz (static linking, drop plugins), the fallback is metacity by default, right?
[17:10] <pitti> I think that'd be acceptable
[17:10] <seb128> yes
[17:10] <pitti> (unlike the "don't render the desktop" one for nautilus)
[17:10] <seb128> but even without compiz and without nautilus and without xsession.d
[17:10] <pitti> ^ I'd really like to avoid that
[17:10] <seb128> we still use some 8 seconds now
[17:10] * rickspencer3 agrees with pitti and seb128 wrt metacity be default
[17:10] <seb128> so that's not enough to rely on it
[17:10] <rickspencer3> seb128, what does it mean "without nautilus"?
[17:11] <seb128> we still need to make progress there
[17:11] <pitti> no, just for prioritization
[17:11] <seb128> rickspencer3, I removed it from the session
[17:11] <pitti> we should start with panel, applets, and nautilus
[17:11] <seb128> no nautilus started at all
[17:11] <seb128> right
[17:11] <rickspencer3> so it doesn't start until you pick somehting from places, for example?
[17:11] <seb128> compiz is a would be nice item but doesn't seem the higher priority
[17:11] <pitti> seb128: also, you have a g-s patch to start everything in parallel, no?
[17:11] <seb128> we have fallback plan there
[17:11] <seb128> rickspencer3, yes
[17:11] <seb128> pitti, yes
[17:11] <seb128> I've included some details and comment on the wiki
[17:12] <seb128> I will try to do that every week with my activity report
[17:12] <seb128> starting everything in parallel makes small difference
[17:12] <seb128> we replace delay by io and cpu bounding
[17:12] <pitti> but gives us more potential for optimization
[17:12] <seb128> yes
[17:12] <seb128> chrisccoulson has been looking at gnome-session
[17:12] <pitti> if panel starts 5 seconds in, we can speed it up as we like and still break the bar
[17:12] <seb128> and gnome-settings-daemon
[17:13] <seb128> there is some wins to do there too
[17:13] <seb128> starting dkpower later
[17:13] <seb128> he will try to optimize gconf reading too
[17:13] <pitti> as a "first against the wall" thing, can we drop some applets? -> proposal: "show desktop", trash, firefox/help launchers
[17:13] <seb128> that's a summary of what we did this week
[17:13] <seb128> I would like to discuss the applet selection
[17:13] <seb128> and maybe changing to 1 gnome-panel bar
[17:14] <pitti> yay
[17:14] * rickspencer3 agrees with dropping all of pittis list
[17:14] <seb128> 1 or 2 bars doesn't make a real speed change
[17:14] * tseliot uses Docky
[17:14] <seb128> but if we reduce applets we could as well fit on one
[17:14] * pitti always disables the bottom bar, not for speed, but for screen real estate reasons
[17:14] <seb128> same here
[17:14] <rickspencer3> pitti, what do you do for window switching?
[17:14] <seb128> I agree with pitti's list
[17:14] <seb128> trash costs 0.5 seconds right now
[17:14] <pitti> rickspencer3: it's in my top panel
[17:14] <seb128> the messaging applet costs 0.5 seconds too
[17:15] <seb128> but I guess we can't go around this one
[17:15] <rickspencer3> hmmm
[17:15] <seb128> we can try pushing dxteam to optimize things though
[17:15] <rickspencer3> that's too long
[17:15] <rickspencer3> kenvandine, any thoughts on loading the MI?
[17:15] <pitti> rickspencer3: http://people.canonical.com/~pitti/photos/pitti-virtual-desktop.png<<BR>> [17:15] <kenvandine> rickspencer3, no...
[17:15] <kenvandine> it starts a bunch of daemons
[17:15] <pitti> (it's from intrepid or so)
[17:15] <kenvandine> maybe look at how that is being done
[17:16] <pitti> indeed all the MI stuff is exceptionally resource intense, with many d-bus services launched, etc.
[17:16] <rickspencer3> does it make sense to start all the daemons at startup?
[17:16] <kenvandine> it would be reasonable to get dx to review that
[17:16] <pitti> there is 6 Info (!) indicator related processes
[17:16] <seb128> I get between 0.5 and 0.8 second win on the test box without the message indicator
[17:16] <rickspencer3> I mean, what can they be listening for if you haven't actually started any apps yet?
[17:16] <seb128> so there is for sure place for doing better
[17:17] <seb128> it's some 15% of the login budget right now
[17:17] <kenvandine> rickspencer3, yeah
[17:17] <pitti> we could perhaps at least defer it
[17:17] <ccheney> oh yea if anyone else needs 10v for this testing they might want to get them soon, i have a feeling they will be EOLd around jan 7
[17:17] <kenvandine> make it on demand
[17:18] <pitti> kenvandine: I guess the daemons listen to some signals, so that they can't be dbus activated when you e. g. open empathy or evo?
[17:18] <kenvandine> yeah
[17:18] <rickspencer3> kenvandine, can you take the action to investigate defering startup of MM related daemons?
[17:18] <kenvandine> yes
[17:18] <seb128> https://blueprints.edge.launchpad.net/ubuntu/+spec/desktop-lucid-startup-speed has a detailed list of work items
[17:18] <seb128> if somebody wants to claim some please do
[17:18] <bryce> ccheney, sucks! Hope they have something similar to replace it.
[17:18] <pitti> kenvandine: there might be a possibility of one process listening to all the signals, and then start the other daemons on demand?
[17:18] <seb128> put your [name] on the line
[17:19] <rickspencer3> ACTION: kenvandine to investigate deferring startup time of MM
[17:19] <ccheney> bryce: i don't know for certain, but pine trail is supposed to be out on that day so very likely will be update to it
[17:19] <seb128> kenvandine, thanks
[17:19] <pitti> so, if we removed above applets, nobody would cry out?
[17:20] <seb128> going back to the applets list
[17:20] <seb128> should we talk to design team?
[17:20] <rickspencer3> seb128, nah
[17:20] <kenvandine> seb128, mind if i add a work item for the indicator services?
[17:20] <seb128> pitti, I'm not sure about the launchers
[17:20] <pitti> also, removing these launchers is a long-term goal anyway, wasn't it?
[17:20] <seb128> I'm all for droping the show desktop and trash ones
[17:20] <rickspencer3> seb128, tbh. I've wanted to get rid of the ff and help launchers anyway
[17:21] <seb128> kenvandine, oh, please do!
[17:21] <pitti> ^ those two sound good indeed
[17:21] <tseliot> bryce: the mini 12 should be available sooner or later
[17:21] <seb128> ok
[17:21] <seb128> should be try to go on one bar then?
[17:21] <pitti> tseliot: does that boot any faster? :-)
[17:21] <seb128> in which case do we want to change the clock applet to just do time by default?
[17:21] <seb128> and not date
[17:21] <pitti> seb128: oh, please
[17:21] <seb128> I'm for keeping the weather though
[17:22] <tseliot> pitti: I don't have it yet. Here's an article about it: http://apcmag.com/scoop_we_review_the_inspiron_mini_12__dells_supersized_yet_superslim_12_inch_netbook.htm<<BR>> [17:22] <pitti> having the date is pretty silly; you get it when hovering over it, and it's the same all day :)
[17:22] <pitti> tseliot: j/k
[17:22] <tseliot> aah
[17:22] <seb128> ok
[17:22] * ccheney thinks the weather is useless after the gweather change last cycle
[17:22] * ccheney just uses a FF weather applet
[17:22] <rickspencer3> seb128, would it speed up desktop time to just load the clock and no evo calendar, weather, etc...?
[17:22] <seb128> so one bar, no launcher and tasks list where those are now, no show desktop, no trash, only time for the clock
[17:22] <pitti> it's not important enough for me to see all the time, but that's just me
[17:22] * pitti has a window
[17:23] <pitti> seb128: and no seconds either
[17:23] <seb128> right, that's not on now
[17:23] <seb128> I didn't plan to change that ;-)
[17:23] <seb128> rickspencer3, I'm not sure, but I will keep doing charts next week and have those for next meeting
[17:23] <seb128> let's do what I said as a first step
[17:23] <seb128> and see for changes over that in next meeting?
[17:23] <pitti> having an extra 25 pixels is a big win by itself
[17:24] <pitti> seb128: sounds good
[17:24] <seb128> we can still tweak some options it's early in the cycle
[17:24] <seb128> anybody having other ideas about things we should change?
[17:25] <pitti> seb128: we'd move the work area switcher to the top bar, too?
[17:25] <seb128> I think we should also try to document how we do this work
[17:25] <seb128> our metrics, tools, etc
[17:25] <rickspencer3> seb128, does wallpaper have an impact?
[17:25] <ccheney> tseliot: mini 12 doesn't work with current ubuntu afaik, has gma500 (ugh)
[17:25] <pitti> I'm also working on app menu caching right now
[17:26] <rickspencer3> other theme stuff?
[17:26] <seb128> pitti, yes, or do you think we should keep the second bar because it's too much?
[17:26] <pitti> seb128: I propose that we dedicate this week to panel (applets, caching), and then revisit next week
[17:26] <seb128> rickspencer3, theme and wallpaper have an impact yes
[17:26] <seb128> I didn't measure how much changing the image format impacts on the loading though
[17:26] <pitti> seb128: no, as I said I'm all for removing the second bar
[17:26] <seb128> would be worth testing if somebody wants to do it
[17:26] <ccheney> if we need more pixels getting rid of the username on right could help a lot
[17:26] <rickspencer3> so simple one color picture for wallpaper, and fast loading theme by default?
[17:27] <pitti> rickspencer3: I guess we won't get that past design
[17:27] <kenvandine> rickspencer3, remember there are plans for that applet
[17:27] <seb128> I've the feeling design team will not like those suggestions ;-)
[17:27] <tseliot> ccheney: it's not the same mini 12 I was talking about. I picked up the wrong link
[17:27] <ccheney> tseliot: ok
[17:27] <rickspencer3> well, we should start with getting the desktop to load within budget
[17:27] <rickspencer3> then we can work on making exceptions, etc...
[17:27] <pitti> /sys/class/dmi/id/sys_vendor == "Dell" -> no background :)
[17:28] <rickspencer3> but design should work within the same constraints that we do
[17:28] <kenvandine> seb128, maybe set group when space is limited in the window list would be a good default for a single panel
[17:28] <seb128> if user == keybuk, no background? ;-)
[17:28] <pitti> kenvandine: do you think we can drop the u1 applet by next week?
[17:28] <kenvandine> no
[17:28] <didrocks> seb128: ahah
[17:28] <kenvandine> well... maybe for lucid... i is early :)
[17:28] <kenvandine> s.i/it
[17:29] <kenvandine> s/i/it
[17:29] <seb128> rickspencer3, I will measure background and theme impact for next week
[17:29] <seb128> trying different formats and plain color
[17:29] <rickspencer3> k
[17:29] <seb128> so we can know how much it costs and if it's worth
[17:29] <seb128> enough on the topic it's almost time
[17:29] <seb128> so just to summarize
[17:29] <seb128> https://blueprints.edge.launchpad.net/ubuntu/+spec/desktop-lucid-startup-speed has work items
[17:30] <seb128> feel free to grab or add some there
[17:30] <seb128> http://people.canonical.com/~seb128/bootchart/ has charts
[17:30] <seb128> I can test on a mini10v so if you want to get changes tested let me know
[17:30] <pitti> so let's do the panel related changes this week
[17:30] <seb128> thanks everybody for helping there ;-)
[17:31] <seb128> I will do the gnome-panel changes this week
[17:31] <pitti> seb128: for next week, it would be great if you could do one with metacity, and the panel changes?
[17:31] <seb128> and measure impact from other applets, theme, background
[17:31] <pitti> (I hope I can get the caching in this week)
[17:31] <seb128> sure
[17:31] * kenvandine plans to upgrade to lucid this afternoon
[17:31] <seb128> I've been doing something similar this morning
[17:31] <rickspencer3> seb128, thanks, great job!
[17:31] * pitti hugs seb128 for coordinating this and his great work
[17:32] <seb128> it was around 9 seconds
[17:32] <seb128> but with the non optimized gnome-session
[17:32] <seb128> so I guess we would be around 8 seconds on the optimized one
[17:32] * seb128 hugs pitti and rickspencer3
[17:32] <seb128> thanks guys
[17:33] <rickspencer3> any other business?
[17:33] <pitti> . o O { gvfs-trash, WTH }
[17:33] <pitti> that belongs started on demand..
[17:33] <seb128> pitti, trash applet?
[17:33] <pitti> seb128: if the applet is gone, so is this?
[17:33] <seb128> pitti, it's delayed until nautilus query for it
[17:34] <pitti> zap it! zap it!
[17:34] <pitti> :)
[17:34] <seb128> ;-)
[17:34] <bryce> sorry, xchat seems to have frozen the last 10 min
[17:34] <bryce> <bryce> seb128, btw the moblin patches are now in lucid. dunno if they helped or if so by how much, but I do notice xkbcomp absent on your chart so guess that's some progress
[17:34] <bryce> seb128, fastest I've gotten X on my hardware is 2 sec, about the same as you
[17:34] <bryce> moblin says they got it to 1 sec, but dunno how (jbarnes wasn't sure either; maybe kernel stuff)
[17:35] <rickspencer3> bryce, in case you were caught in the same time warp as ccheney, it's 9:34am our time (unless I am also in a time warp)
[17:35] <pitti> bryce: is that with udev already? or hal?
[17:35] <bryce> pitti, nope neither
[17:35] <seb128> bryce, yes, your changes won us almost 1 second, thanks ;-)
[17:35] <pitti> bryce: heh, xorg.conf?
[17:35] <rickspencer3> bryce, has keybuk confirmed that mobile *really* got x to start in 1 second?
[17:35] <bryce> pitti, don't think so
[17:36] <rickspencer3> like, is that from a boot chart that we made, or is this just something they told us?
[17:36] <pitti> bryce: no input devices at all then? :)
[17:36] <ccheney> iirc he said he had and that it was something in the binary itself
[17:36] <bryce> rickspencer3, it is from what they told us
[17:36] <rickspencer3> ok
[17:36] <ccheney> but i am a bit fuzzy on the details
[17:36] <bryce> rickspencer3, (according to keybuk)
[17:36] <rickspencer3> oops, ok, I'll ask him later
[17:37] <bryce> pitti, I mean "without the hal removal changes"
[17:37] <pitti> bryce: ah, ok; perhaps that'll win another .1 seconds
[17:37] <seb128> oh
[17:37] <seb128> dropping compiz-fusion-plugins-extra wins 0.5 seconds
[17:37] <seb128> we don't use it anyway
[17:37] <seb128> so it should go to universe soon
[17:38] <tseliot> bryce: are you using the kernel patches that I recommended already (to reduce X startup time)?
[17:39] <seb128> rickspencer3, I think that's a "no other business"
[17:39] <rickspencer3> lol
[17:39] <seb128> ie those are after meeting discussions
[17:39] <rickspencer3> okay then
[17:39] <seb128> (in case you want to official wrap the meeting now)
[17:39] <rickspencer3> thanks all
[17:39] <seb128> thanks rickspencer3
[17:39] * rickspencer3 taps gavel

CategoryDesktopTeam

Back to DesktopTeam.

DesktopTeam/Meeting/2009-12-01 (last edited 2009-12-02 01:44:52 by c-98-200-151-46)