=== IdleOne is now known as Guest21295 === Guest21295 is now known as IdleOne [17:02] hi Renato! ;) [17:03] hi Albert :) [17:03] hey Zoltan [17:05] Saviq: you greeted me while i wasn't here yet [17:06] tsdgeos, that's because I heard you on Icecast :) [17:06] did you? [17:06] ok [17:06] stop looking at the screen jonathan [17:07] hello [17:07] hi [17:07] hello world [17:07] hey [17:07] #ubuntu-uds-junior-ballroom-1 (for those who can't read the shortened version on screen) [17:08] Hello Saviq [17:08] I wonder who can kick udsbotu to update the title [17:08] yes we can [17:08] We can hear you on the stream. [17:09] http://www.macieira.org/blog/qt-stats/ [17:10] o/ [17:12] just saying hallo [17:16] hey pstolowski [17:16] http://qt-project.org/wiki/QtMediaHub [17:18] AlanBell, o/ [17:19] in Ubuntu with QML (a11y backported into 4.8 by fregl), Unity-2D is said to be "99% accessible" [17:19] AFAIK the same went into Qt5 [17:20] I would like to see A11Y be moved lower in the stack anyway (e.g., wayland), and then, as thiago said, it would be in the framework bits of Qt, without app developers having to worry about it too much. [17:21] mikhas, not that easy, just reading the text of, say, a button, is often not enough [17:21] mikhas, there needs to be additional data about things that are represented visually [17:22] no, it isnt, you need screen reading, UI navigation and input. [17:22] (at least) [17:22] Saviq, hey [17:22] simple screen reading could be implemented lower, but an app developer still would need to be able to influence it, or replace, even [17:23] Saviq, I talked to fregl several times, and to my understanding Qt5 has a11y reworked compared to 4.8+a11 patch [17:23] mikhas, think the launcher icons, we have data like app name, how many windows are running, progress bars, notification count [17:23] pstolowski, API wise, or just backend wise? [17:23] Saviq, e.g. in 4.8+patch there are still areas that just have stub/unimplemented code (e.g. text input in qml) [17:24] pstolowski, yeah, sure, so it got even better in Qt5 [17:24] Saviq, that's a good reason why it needs to lower in the stack. [17:24] Saviq, afaik public api stays, but underneath it has been reworked and cannot even be backported easily anymore [17:25] Imagine Unity running on Wayland, then you'd have all the bits in one place. [17:25] http://wiki.qt-project.org/Devices pi packages - http://wiki.qt-project.org/Devices/RaspberryPi [17:25] mikhas, all that data needs to be presented to a11y, you don't want to navigate from icon to progress bar to notification counter etc. [17:26] I know, there has been some great research by http://www.aegis-project.eu/ to find out more about that, and to come up for solutions. [17:26] *with solutions [17:30] re: Wayland + runtime switch; a good example of why frameworks can be useful ;-) [17:38] what about functional testing? testability seems to be orphaned? what's the recommended UI testing platform going forward? [17:41] Saviq, there is TDriver right now, for UI testing, but it uses Ruby as its scripting language … [17:41] mikhas, yes, and AFAIK it's being orphaned i.e. noone sponsors it anymore [17:41] (TDriver probes into Qt UIs through A11Y layer, AFAIK) [17:42] not a11y [17:42] it introspects the object tree [17:45] gtg === daker is now known as daker__ === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Track: Cloud & Server | OpenStack HA | Audio: === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Track: Cloud & Server | OpenStack HA | Audio: http://icecast.ubuntu.com:8000/junior-ballroom-1.ogg.m3u === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Track: Cloud & Server | OpenStack HA | Url: | Audio: http://icecast.ubuntu.com:8000/junior-ballroom-1.ogg.m3u === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Track: Cloud & Server | OpenStack HA | Url: http://summit.ubuntu.com/uds-q/meeting/20674/servercloud-q-openstack-ha/ | Audio: http://icecast.ubuntu.com:8000/junior-ballroom-1.ogg.m3u [18:34] FYI: Parallel discussion regarding "cloud agnostic way to define resources in juju" in room 201 .. OpenStack interested folks might want to join in. [18:40] We could use a glance expert in room 201 actually, trying to figure out how much glance exposes about what OS an image provides. [18:50] uds-junior-ballroom-1: 5 minutes left in this session! [18:51] uds-junior-ballroom-1: 4 minutes left in this session! [18:52] uds-junior-ballroom-1: 3 minutes left in this session! [18:53] uds-junior-ballroom-1: 2 minutes left in this session! [18:54] uds-junior-ballroom-1: 1 minute left in this session! [18:55] uds-junior-ballroom-1: This session has ended. === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-q/junior-ballroom-1/ - http://ubottu.com/uds-logs/%23ubuntu-uds-junior-ballroom-1.log [19:00] hello [19:02] anyone in the room? [19:04] hmm === zyga_ is now known as zyga === oSoMoN_ is now known as oSoMoN === ChanServ changed the topic of #ubuntu-uds-junior-ballroom-1 to: Track: Cloud & Server | Openstack Juju Charms Next Steps | Url: http://summit.ubuntu.com/uds-q/meeting/20673/servercloud-q-openstack-charms/ | Audio: http://icecast.ubuntu.com:8000/junior-ballroom-1.ogg.m3u [23:16] hallohallo [23:16] o/ reading you [23:16] you are on screen [23:16] how was the audio check? [23:17] * med_ waves at rnathuji [23:17] I'm not able to get audio [23:18] Anyone else having audio problems? [23:18] http://icecast.ubuntu.com:8000/junior-ballroom-1.ogg direct link.. works great for me. [23:18] got it now [23:19] cool. [23:19] adam_g is talking through the etherpad. [23:19] awsome? [23:21] adam_g: The CI lab charms should be the ones that are in lp:charms/* ... is that the case? === zyga_ is now known as zyga [23:22] Its no small feature. Its huge! [23:24] SpamapS: It's almost awesome. [23:24] It allows hooks to defer their actions until they can succeed, so charms don't need to be dependent on the order relations are established. [23:26] I have one idea for that using 'charm splice' [23:27] can you then kill the existing spliced charm once other nodes providing that service are up? [23:28] subordinates are done, but they carry a 1:1 relationship for all components [23:29] I/O death [23:30] building 7*mini-ubuntu's are painful for I/O [23:30] agreed ttx .. splice is a hack [23:30] * SpamapS said the Hacker [23:30] heh. [23:32] Should work on RAX or HP Cloud if we can solve the S3 issues. [23:32] its not anywhere official yet because its fairly hacky [23:33] To be clear, for basic dev purposes, you can use the 'placement: local' hack [23:33] lp:~clint-fewbar/charm-tools/add-splice-command <-- the hack [23:33] "placement: local" is the LXC local provider? Or something else? [23:34] med_: placement: local causes services to be deployed to node 0 [23:34] in containers? [23:34] LXC has the ability to do copy-on-write for the 'lxc-clone' already [23:34] that would help a lot [23:35] There were also issues with scsi namespacing [23:35] perhaps somebody could go find smoser so he can comment? [23:36] yes [23:36] The CI lab charms should be the ones that are in lp:charms/* ... is that the case? [23:37] drbd should be doable as a subordinate [23:38] pacemaker needs a reliable network between nodes does it not? [23:38] SURE! [23:38] * SpamapS subordinates [23:39] I just wonder about how well we can define "a dedicated network between those nodes" w/o tagging in maas [23:39] drbd really needs the same thing [23:40] I want all the goodness in the main charms. No me gusta forks. [23:40] You should make your hacks subordinates. [23:40] Or config.yaml items [23:41] subordinates ? [23:41] I want people to be able to use the charm store to be able to repeat the CI testing. It will gather effort around CI testing of openstack around the same place rather than encouraging it to diverge. [23:42] ttx: subordinates can be used to add custom functionality to a generic charm. [23:42] rnathuji, anything you wanted to bring up in the charm discussions? [23:42] Alright I trust you to give us all/most of the important stuff :) [23:42] Thats all I have. [23:45] no, nothing specific. it sounds like things are in a state to try throwing onto arm clusters w/lxc as virt. I haven't yet done that exercise, though :). I'm also curious as to quantum support, but that sounds nascent yet. [23:45] ++++ [23:46] session tomorrow about openstack+ceph/rbd [23:47] thanks adam_g [23:47] CEPH upstream will be in attendance [23:47] ceph charm sucks [23:47] * SpamapS said the author [23:47] cheers! [23:51] ttx: I hear you, and I agree. Make a juju based devstack so real deployments are more like developer deployments thus eliminating disconnect bugs! [23:52] ___________________________________ [23:52] < somebody tell ttx to look up here > [23:52] ----------------------------------- [23:52] \ ^__^ [23:52] \ (oo)\_______ [23:52] (__)\ )\/\ [23:52] ||----w | [23:52] || || [23:53] corvus: you sound great ;) [23:54] 10 seconds? [23:54] yes [23:54] :( I hate living in the past [23:55] so true [23:55] uds-jb-1: 5 minutes left in this session! [23:56] uds-jb-1: 4 minutes left in this session! [23:57] uds-jb-1: 3 minutes left in this session! [23:58] uds-jb-1: 2 minutes left in this session! [23:58] juju set mysql dataset-size=8M <-- tunes mysql to use almost no RAM [23:59] uds-jb-1: 1 minute left in this session! [23:59] juju set mysql tuning-level=fast <--- tunes mysql to not use fsyncs much [23:59] __________________ [23:59] < jim look up here > [23:59] ------------------ [23:59] \ ^__^ [23:59] \ (oo)\_______ [23:59] (__)\ )\/\ [23:59] ||----w | [23:59] || ||