[13:50] <ev> whoop
[13:54] <xnox> ev: that's a bug, here is patch: -whoop \n +whoopsie
[13:55] <ev> :D
[13:55] <ev> xnox: mind signing the CLA and then resubmitting?
[13:55]  * xnox part
[13:55] <ev> lol
[13:55] <xnox> =)
[13:59] <slangasek> ev (et al): https://plus.google.com/hangouts/_/89aaf5a761d6e92baa1fcb798e6adf406b11c194
[14:01] <cjwatson> ev: your audio is rather glitchy
[14:01] <pitti> is that just me, or is ev really hard to understand?
[14:01] <sergiusens> pitti: it is
[14:02] <sergiusens> ev: or turn off video
[14:02] <sergiusens> yeah
[14:02] <cjwatson> ev: flashblock is really helpful so that when you open the summit page it doesn't autoplay the video
[14:05] <tedg> Could we have tools to build a symbol table that they could upload as well?
[14:06]  * tedg is probably on delay :-)
[14:06] <sergiusens> like breakpad/socorro?
[14:06] <ev> breakpad and socorro are something different
[14:07] <ev> fedora's darkserver would be the closest existing match, as far as I've gathered so far
[14:07] <ev> for a symbol server that is
[14:08] <sergiusens> right
[14:11] <pitti> sorry, ffox crashed
[14:11] <ev> you're still visible
[14:11] <ev> err were :)
[14:11] <xnox> pitti: rodger
[14:36] <dobey> did the stream blow up or something?
[14:37] <fugue88> dobey: Still working for me so far.
[14:38] <dobey> oh now it's back
[14:38] <dobey> it froze up for me around 33:05 :-/
[14:39] <xnox> ev: on android maps for examples, one can shake the phone in frustration - and a popup appears: What's wrong? [] navigation wrong [] wrong location [] missing information & screenshot. And you can submit feedback.
[14:39] <sergiusens> what about bugs (not crashes) for click packages that belong to ubuntu?
[14:39] <xnox> ev: how would we centrally deal with people reporting bugs (non-crashes) in reviews on the app store?
[14:40] <xnox> !!!! =(
[14:40] <sergiusens> cjwatson: what about during the development phase?
[14:40] <cjwatson> xnox: I think that's specific to maps
[14:41] <cjwatson> sergiusens: even click packages that belong to Ubuntu don't have an existence in Launchpad that we can automatically discover, at present - no equivalent of /ubuntu/+source/packagename
[14:41] <slangasek> xnox: why should those be dealt with centrally?  if people are panning the app in the reviews, that's the app dev's problem
[14:41] <cjwatson> I don't really think we should have a concept of "click packages that belong to Ubuntu"
[14:41] <cjwatson> they happen to be in the com.ubuntu namespace - so what
[14:41] <sergiusens> cjwatson: sounds good, I'm just looking for a standard
[14:41] <cjwatson> we shouldn't build something just for that namespace
[14:42] <slangasek> cjwatson: does that mean we don't have a way to receive bug reports from users about them?
[14:42] <xnox> slangasek: cjwatson: I hoped that it should be easy to (a) file review vs (b) file report. In app store. I guess we will see, those who care, to start providing feedback buttons within their app.
[14:42] <sergiusens> I'm more worried about QA
[14:43] <sergiusens> pitti: that's what we've been doing since we lived on PPAs
[14:43] <pitti> sergiusens: "that" == ?
[14:44] <tedg> I guess perhaps we could "make it easy" if they use LP, but allow other things as well.
[14:44] <sergiusens> pitti: sorry, audio delay that == use launchpad project
[14:44] <pitti> sergiusens: well, you need an LP account and project to build your branch in a PPA
[14:44] <pitti> sergiusens: that's not true for ClickPkgs?
[14:44] <cjwatson> it is not
[14:44] <pitti> tedg: yes
[14:44] <cjwatson> the app store does not require an LP account
[14:44] <xnox> cjwatson: after all Android Open Source project exists to file bugs against "core ubuntu apps"
[14:45] <sergiusens> pitti: sorry, I think I incorrectly phrased what my sentence
[14:45] <cjwatson> I'm not sure I see the connection between the AOSP and our core apps
[14:45] <xnox> cjwatson: and then phone-app will have for example https://bugs.launchpad.net/phone-apps/+filebug ?!
[14:46] <cjwatson> that's one possibility
[14:46] <cjwatson> (I'm not sure why the interrobang.  Is it so surprising?)
[14:46] <sergiusens> host side
[14:46] <xnox> slangasek: that URL should be in the app store next to the app.
[14:46] <sergiusens> ev: design bugs perhaps?
[14:47] <sergiusens> messaging-app messages not getting to the dialer-app properly
[14:50] <xnox> slangasek: i can rewind the video and play back again =)
[14:50] <slangasek> hah, true
[14:52] <ev> apols for my thinkpad being rubbish
[15:04] <danjared> Can someone provide me with the hangout URL?
[15:05]  * danjared waves at superm1 
[15:05] <superm1> hi
[15:05] <superm1> is the video active?
[15:05] <danjared> not sure. I need the link to the hangout
[15:05] <slangasek> sorry, just a minute behind
[15:05] <danjared> k
[15:06] <superm1> danjared: i'm not convinced that hangouts will work on the network here.
[15:06] <danjared> superm1: I'm at home
[15:06] <superm1> ah
[15:06] <danjared> superm1: use the guest wireless?
[15:06] <superm1> yeah i might have to
[15:06] <slangasek> https://plus.google.com/hangouts/_/a0c663c24c1389a17669272204fdc82dc71e259b
[15:07] <slangasek> superm1: would it be better if we called you?
[15:08] <superm1> maybe, give me a sec
[15:09] <slangasek> superm1: hmm, it doesn't give me the option to add PSTN
[15:09] <superm1> well go ahead and get started, danjared can speak up and i'll throw in text responses if need be here
[15:11] <superm1> i'm hearing echo like nuts
[15:12] <cjwatson> if you have the summit page loaded then make sure it isn't playing the video too
[15:12] <superm1> oh that's it
[15:12] <cjwatson> flashblock is helpful for this
[15:14] <zyga> QUESTION: does that include patching command line tools like pm-hibernate?
[15:16] <danjared> zyga: yes, we'd want to patch that
[15:17] <zyga> is that MBR, GPT or either?
[15:17] <zyga> (the magic partition that is required)
[15:17] <danjared> either are supported
[15:17] <danjared> "Dedicated partition == to size of memory type 0x84 in MBR layout or GUID D3BFE2DE-3DAF-11DF-BA-40-E3A556D89593 in GPT layout."
[15:18] <zyga> can we somehow double that as hibernate swap file and always enable it
[15:19] <danjared> double the partition as a swap file?
[15:19] <zyga> well, swap partition
[15:19] <zyga> to create that partition always
[15:19] <zyga> but use it as a hibernate partition / swap partition if the feature is not available
[15:20] <danjared> no, that's not possible
[15:20] <danjared> it's a special partition with a proprietary layout
[15:20] <slangasek> well
[15:20] <xnox> zyga: well if it is mounted as swap, one can't rapid hibernate to it.
[15:21] <slangasek> that doesn't prohibit the kernel from reusing the partition once it's determined that irst isn't supported
[15:21] <danjared> correct
[15:21] <danjared> you just can't reuse it while the firmware is using it
[15:21] <slangasek> right
[15:22] <danjared> I'm not too good at multitasking irc/video :)
[15:22] <zyga> smart!
[15:32] <zyga> is that something that's safe to enable (say, on the plane)
[15:33] <zyga> when the laptop wakes up to hibernate and hangs
[15:34] <zyga> fair enough
[15:34] <zyga> not our fault
[15:37] <slangasek> zyga: sure, it's perfectly safe, just not legal ;)
[15:37] <cking> i guess one should drop caches before going into hybrid sleep
[15:38] <xnox> zyga: it doesn't wake up to write out ram to ssd.
[15:38] <zyga> expanding memory is not something that most users will do, I think we should optimize on SSD usage instead
[15:38] <zyga> xnox: ah, I didn't know that
[15:39] <zyga> slangasek: why is it not legal?
[15:39] <xnox> zyga: any electrical activity causes EMC waves, in practice it will be ambient noise that is accounted for & EMC protected in the cockpit.
[15:40] <zyga> ah
[15:41] <zyga> does intel rapid storage supports encryption?
[15:42] <zyga> or some other full-disk-encryption feature that's available at firmware level
[15:42] <cjwatson> that's what we're talking about with the ATA encryption extension
[15:42] <cjwatson> whatever it's actually called :)
[15:42] <zyga> right, but does the rapid storage thing itself support that?
[15:44] <danjared> zyga: confusingly, I'll also mention that Intel Rapid Storage Technology ("RST") is different from Intel Rapid Start Technology ("iRST")
[15:44] <zyga> oh
[15:44] <zyga> sorry, I meant rapid start
[15:44] <zyga> rapid storage is probably some form of firmware raid
[15:45] <danjared> correct
[15:45] <zyga> can we just zero-out unused memory instead of doing that via special interface?
[15:45] <danjared> formerly called Intel Matrix
[15:46] <danjared> zyga: that's what we suspect needs to be done
[15:46] <cjwatson> zyga: you still have to spend time writing the zeroes if it doesn't know that they're zero for a reason, surely
[15:46] <cjwatson> .oO( just write the 1 bits, should be about twice as quick )
[15:46] <zyga> cjwatson: well, it should be fast enough, unless it's something that needs to be done all the time, not just before the actual 'suspend'
[15:47] <cjwatson> zyga: but it gains you nothing over just writing out the unzeroed memory
[15:47] <cjwatson> you only get a performance gain if you can not write it at all
[15:47] <cjwatson> if you see what I mean
[15:47] <xnox> danjared: well "Intel Matrix" marketing at one point included intel raid as well.
[15:48] <zyga> cjwatson: right, I assume that the firmware somehow skips zeros (probably does some form of RLE compression)
[15:48] <danjared> xnox: yeah, RST formerly Matrix is fakeraid
[15:48] <cjwatson> It would be nice to know whether that assumption is true
[15:49] <zyga> cjwatson: can we fill the partition with /dev/urandom, copy it somewhere, suspend with half the memory zeroed and check if the firmware zero-erased the whole suspend partition?
[15:49] <zyga> (on resume0
[15:49] <cjwatson> perhaps
[15:49] <zyga> if it actually zeroes-out all 8GB each time then it's a pretty useless feature to do that anyway (optimize)
[15:50] <zyga> dbus signal!
[15:56] <zyga> any testing things that we need to have in checkbox?
[15:56] <zyga> like
[15:56] <zyga> can we check if we resumed form rapid start?
[15:56] <zyga> vs from some regular type of resume (or just normal boot)
[15:57] <zyga> cjwatson: ^^
[15:59] <zyga> cjwatson: I also have a bit of hardware that has support for that
[16:00] <danjared> zyga: it's supposed to be completely transparent to the system, so probably not unless Intel has some super secret way of telling the OS about that
[16:00] <zyga> slangasek: ^^ (question about resume detection)
[16:00] <zyga> ah
[16:00] <zyga> too bad
[16:01] <slangasek> right, the OS isn't meant to know anything about it :)
[16:01] <xnox> zyga: it's a pure firmware feature, thus the tests could be if the: detection/check in partman correctly detects if the system has/hasn't rapidstart without rapid partition present. Such coverage info would be useful for cjwatson.
[16:01] <xnox> zyga: in terms of feature, sleep & resume should work =)
[16:01] <danjared> :)
[16:02] <slangasek> danjared: are you familiar with the blueprint drafting process? getting work items recorded etc?  (you're marked as drafter on the blueprint)
[16:02] <danjared> not terribly
[16:02] <slangasek> ok
[16:02] <slangasek> no time like the present for learning ;)
[16:02] <danjared> I just added modifications to superm1's first draft
[16:03] <slangasek> so the main things are: 1) get the data copied out of the etherpad to somewhere more permanent, 2) get the work items captured in the "work items" section of the blueprint in the prescribed manner, so that they get automatically tracked
[16:04] <danjared> oops, I forgot that there was even an etherpad, and I'm having trouble logging in to it
[16:04] <slangasek> https://plus.google.com/hangouts/_/b7bacb6e4c7d2466cbac7a06a9813080d9debfdc
[16:05] <ogra_> ChickenCutlass, ^^^
[16:06] <danjared> slangasek: I'm trying to log into pad.ubuntu.com to copy the etherpad contents but am in an endless "you don't have access" -> "log in" -> "you don't have access" loop
[16:06] <slangasek> danjared: the login certainly works better if you do it in a separate window: http://pad.ubuntu.com/uds-1308-foundations-irst-support
[16:06] <xnox> danjared: yeah, you need to be member of https://launchpad.net/~ubuntu-etherpad
[16:07] <danjared> ah
[16:07] <rsalveti> http://pad.ubuntu.com/uds-1308-foundations-1308-power-measurement
[16:07] <xnox> danjared: join the open team https://launchpad.net/~ubuntu-testing which will gain you access to ~ubuntu-etherpad =)
[16:07] <xnox> ... probably a bug in launchpad ACL.....
[16:08] <danjared> now waiting for ACLs to sync, I presume
[16:08] <danjared> if only we had solved single-sign on in the 80's
[16:09] <xnox> danjared: or encrypted hibernate.... =)
[16:09] <xnox>      ^ rapid
[16:09] <danjared> I don't think we even had hibernate in the 80's. we certainly had kerberos then
[16:10] <doanac`> rsalveti: shall i join the hangout?
[16:10] <rsalveti> doanac`: please
[16:10] <rsalveti> doanac`: https://plus.google.com/hangouts/_/b7bacb6e4c7d2466cbac7a06a9813080d9debfdc
[16:11] <slangasek> kerberos \o/
[16:12] <cjohnston> You would have to set thresholds
[16:13] <cjohnston> So that the dashboard can see if it's above or below the threshold
[16:13] <cjohnston> cking: ^
[16:15] <cjohnston> That sounds more like it should go on the power tab and not in the daily smoke testing
[16:16] <gema> cjohnston: you may want to join the conversation :?
[16:16] <cjohnston> want is strong
[16:16] <gema> cjohnston: x)
[16:16] <gema> cjohnston: you have the dash perspective
[16:26] <chiluk> can you guys enable lower third tags?
[16:27] <cking> i also think for long duration runs we could also be using http://kernel.ubuntu.com/~cking/suspend-blocker
[16:27] <cjohnston> hangout toolbox isnt working for me :-/
[16:27] <cking> me too
[16:28] <ogra_> must be a british thing :)
[16:28] <gema> cjohnston: you are british now
[16:28] <ogra_> :)
[16:29] <slangasek> the United Kingdom of Great Britain, Northern Ireland, and Florida
[16:29] <gema> indeed
[16:30] <plars-vuds> I don't think my audio is getting through
[16:30] <plars-vuds> and I'm only getting pieces
[16:30] <ogra_> i dont think your video is either
[16:30] <plars-vuds> we need monkey for ubuntu touch :)
[16:30] <plars-vuds> where can we see this data?
[16:30] <plars-vuds> from powerd
[16:30] <ogra_> sforshee, ^^^
[16:32] <sforshee> plars-vuds: run 'powerd-cli stats'
[16:35] <plars-vuds> can't seem to get my hangout connection back :(
[16:35] <ogra_> plars-vuds, you seem still connected on the screen for me
[16:36] <plars-vuds> ogra_: heh, I'm not :)
[16:36] <sergiusens> rsalveti: the lies!
[16:36] <sergiusens> but no questions :-)
[16:36] <rsalveti> sergiusens: haha :-)
[16:37] <rsalveti> sergiusens: you should just join in
[16:37] <rsalveti> but we're done already
[16:37] <sergiusens> rsalveti: I didn't have the link
[16:37] <sergiusens> didn't want to disturb
[16:37] <cjohnston> cking: https://code.google.com/p/hangout-toolbox/issues/detail?id=17
[16:38] <plars-vuds> sforshee: I'm not sure that powerd-cli stats is showing us anything all that usable... I've tried playing videos, web browser, etc, but I don't see much. However running powerd-cli test seems to bring a lot more things into view
[16:39] <plars-vuds> I just mean that even if we go run a bunch of apps in a system test, and then collect the stats with this, is it basically just going to show us the display request and activity timer?
[16:40] <sforshee> plars-vuds: the reason is that not much is using powerd requests yet
[16:41] <sforshee> plars-vuds: that will change
[16:42] <plars-vuds> sforshee: ah, ok cool
[16:43] <plars-vuds> sforshee: what do all the built-in tests do? Anything device specific? Would it be useful to add them to the daily smoke tests?
[16:43] <sforshee> plars-vuds: they basically verify that the dbus API works as expected
[16:43] <plars-vuds> ok
[16:44] <plars-vuds> so probably not then
[16:44] <sforshee> perhaps not, they shouldn't be change much at this point
[17:56] <mandel> I'm suppose to be in the next talk, how do I join
[17:57]  * mandel is utter useless with this gootle talk thingy
[18:00] <slangasek> https://plus.google.com/hangouts/_/6dd72c850c11015d570e57ed45e02492a2755a11
[18:08] <xnox> Are we live yet?
[18:08] <slangasek> in 5 sec
[18:09] <GuidoPallemans> yup, live now
[18:10] <cjwatson> use flashblock so the youtube stream doesn't automatically start
[18:14] <cjwatson> That was the security team, fwiw, not foundations
[18:14] <slangasek> apparently security is now a subset of foundations, that's fine ;)
[18:14] <jdstrand> ?
[18:14]  * jdstrand is not in this session-- is there something I should address?
[18:15] <lool> notes in http://pad.ubuntu.com/uds-1308-foundations-s-touch-download-service
[18:15] <aquarius> the slangasek empire grows ;)
[18:15] <aquarius> if the interface to the downloader is D-Bus, and my app requests a download and then is later suspended or killed, it may miss any signal which is sent on the download being complete. How does my app, when it gets started up/unsuspended, know whether its downloads were done? It can check if there's a file at the path that it asked for, but how does it know if that download is still working? Will the
[18:15] <aquarius> downloader service remember downloads which have snce completed?
[18:15] <lool> jdstrand: I think Manuel is presenting the security implementation
[18:15] <lool> jdstrand: seems good
[18:16] <jdstrand> ah right, yes, we discussed that a little while ago
[18:16] <slangasek> jdstrand: nah, mandel just seems to have the impression that Foundations is all powerful :-)
[18:16] <jdstrand> :)
[18:17] <GuidoPallemans> QUESTION so in 2 days it will be possible for a QML application to download files? is there an API to use?
[18:18] <aquarius> ah, cool, so the downloader will remember downloads that happened even after they're completed. How does it know when to garbage collect that data? Does an app have to call a OKYouCanDeleteThatDataNowIHaveReadIt api?
[18:19] <lool> aquarius: (want to join?)
[18:19] <lool> aquarius: https://plus.google.com/hangouts/_/6dd72c850c11015d570e57ed45e02492a2755a11
[18:19] <lool> in case you do
[18:19] <aquarius> lool, can't, I'm in meetings :) Just trying to pay some attention
[18:20] <GuidoPallemans> ok, thanks
[18:21] <tedg> Is the download service verifying the app id somehow?
[18:22] <tedg> Or just trusting the App?  ;-)
[18:22] <jdstrand> cjwatson: no
[18:22] <jdstrand> cjwatson: it is a subset of the appid
[18:22] <aquarius> lool, but the downloader service remembersa download, and keeps its D-Bus object path alive, *even after the download is completed* (so my app can ask about that download even after it completed because I might have been suspended when it was completed). How long does the downloader keep that D-Bus object path available for? Forever?
[18:22] <jdstrand> cjwatson: it is XDG_.../$appname
[18:22] <cjwatson> just appname, not packagename/appname?
[18:22] <jdstrand> cjwatson: so it isn't $pkgname_$name_$version
[18:23] <cjwatson> the appname is not unique
[18:23] <tedg> jdstrand, packagename?
[18:23] <barry> what does it use instead of xdg for the system bus service?
[18:23] <jdstrand> let me cehck something
[18:23] <jdstrand> we might have just dropped version
[18:23] <aquarius> QUESTION: is there some system-level process for visualising downloads in progress? An indicator or similar? Or is it up to each app individually to show the progress of its downloads?
[18:23] <cjwatson> just dropping version seems sensible
[18:23] <tedg> aquarius, I imagine they'll go into the sync menu, but that's not for 13.10
[18:24] <jdstrand> ah right
[18:24] <jdstrand> not appname, pkgname
[18:24] <jdstrand> @{APP_PKGNAME}="com.ubuntu.developer.jdstrand.evilapp"
[18:24] <jdstrand>   owner @{HOME}/.cache/@{APP_PKGNAME}/                  rw,      # XDG_CACHE_HOME
[18:24] <aquarius> tedg, yeah, that's what I thought, I was just wondering whether that's actually the plan of record or whether it's just speculation ;)
[18:24] <jdstrand>   owner @{HOME}/.cache/@{APP_PKGNAME}/**                mrwkl,
[18:24] <aquarius> also, like GuidoPallemans, I'd really like to see this exposed to QML. Bindings for it, you know it makes sense ;)
[18:24] <jdstrand> ...
[18:24] <cjwatson> ok, makes sense for different apps in the same package to be able to communicate etc.
[18:24] <jdstrand> cjwatson: ^
[18:25] <jdstrand> cjwatson: exactly. I just forgot for a moment :)
[18:25] <tedg> aquarius, I don't know of a plan of record there... we should probably make one.  But I think everyone is thinking "tomorrow" on it right now ;-)
[18:25] <aquarius> I am confused about the discussion of XDG stuff. My app gets its own private directory to save files into: why would the downloader not download files into that folder? (Obviously that folder needs some sort of naming, which will probably be to do with app name, but that's not a downloader-specific problem; everything needs to know where that folder is)
[18:26] <jdstrand> we do use set the APP_ID_DBUS to be versioned though
[18:26] <jdstrand> (that is used in the dbus api)
[18:26] <tedg> aquarius, It does get that folder, the private folder is based on the XDG dir.
[18:26] <jdstrand> @{APP_ID_DBUS}="com_2eubuntu_2edeveloper_2ejdstrand_2eevilapp_5fevilapp_5f0_2e5"
[18:26] <aquarius> barry, yeah, that's the OKYouCanDeleteThatDataNowIHaveReadIt API I mentioned above ;)
[18:26] <tedg> jdstrand, Hah!  Good catch!  :-)
[18:27] <barry> aquarius: yep! :)
[18:27] <aquarius> tedg, right... so I'm confused why this is being discussed in the downloader session, since everything needs to know about the name of that folder, and the downloader can just get the name of it from the same place that everything else does :)
[18:27] <tedg> barry, You don't get a dbus signal, but you can add a click hook to run as the user.
[18:28] <tedg> aquarius, Well, we don't have a place "to get it" -- it's "well known" ;-)
[18:28] <GuidoPallemans> but what if I make an app like jdownloader, will its files be automatically deleted after deinstallation? because I probably need the files after deinstallation too
[18:28] <barry> tedg: nod
[18:28] <cjwatson> click hook as in a hook belonging to a system package which is triggered when a click package is installed or removed
[18:28] <jdstrand> aquarius: I'm confused by your question. your private directory is a subdir under the XDG dir
[18:28] <cjwatson> specifically not as in a hook belonging to a click package
[18:28] <aquarius> jdstrand, right, tedg's explained, I get it now :)
[18:29] <jdstrand> aquarius: ok :)
[18:29] <tedg> cjwatson, I think that this is the download-service defining a hook
[18:29] <tedg> cjwatson, So it can clean up its own DB on remove
[18:29] <aquarius> GuidoPallemans, an app can only write files for itself to read. If you download a file then it only belongs to that app. If other apps need it then you have to share the file with those other apps using the content service.
[18:29] <aquarius> cjwatson, is there spec in the hook definition to define a hook which runs on *every* package? Rather than just "packages with file N in them"?
[18:30] <aquarius> cjwatson, I suppose you could make your "call me if this file exists" declaration be "manifest.json" ;)
[18:30] <tedg> GuidoPallemans, How are you getting the App ID?  From Apparmor?
[18:30] <slangasek> (rubbish, not to be confused with the garbage that we collect)
[18:30] <cjwatson> aquarius: *cough* onthetodolist
[18:30] <GuidoPallemans> tedg: I don't know man, I'm just thinking hypothetically
[18:31] <aquarius> cjwatson, I wondered if it might be :P
[18:31] <tedg> I'm more asking because we need to do that for other things as well.
[18:31] <tedg> So thinking we should be consistent :-)
[18:31] <tedg> jdstrand, Is there an easy way to turn a PID into an AppID?
[18:32] <cjwatson> it would perhaps help if we had consistent terminology, since we're seeing people using appid to refer to things that aren't appids
[18:32] <cjwatson> since we made up that term more or less on the spur of the moment, somebody could probably come up with a couple of others :)
[18:32] <jdstrand> tedg: man aa_getcon
[18:33] <jdstrand> cjwatson: yes, that has been a problem for some time :)
[18:33] <cjwatson> md5 - I hope this is just for a basic transport integrity check, not for security?
[18:33] <tedg> jdstrand, Cool!
[18:33] <tedg> cjwatson, +1
[18:34] <cjwatson> I have to step over to the CI/click session.  Drop a line here if something needs me
[18:34] <lool> cjwatson: ok thanks
[18:34] <aquarius> QUESTION: what does the downloader do if it's asked to download, say, http://example.com/thing.txt, and $MYAPPFOLDER/thing.txt already exists? Does the downloader refuse to download it, silently overwrite it, or write $MYAPPFOLDER/thing (1).txt (in which case it'll need an API to say "this is the location I downloaded it to" -- maybe that's needed anyway?)
[18:36] <cjwatson> aquarius: perhaps it would be sensible to download to a hash rather than to the basename of the URL
[18:36] <tedg> barry, Could you install it in the system (both bad and good)
[18:36] <xnox> barry: i thought you don't want to trust "all/system" certificates, and only trust the update SSL certs.
[18:37] <barry> tedg: i'd really like to avoid mucking with system global state for a test
[18:37] <tedg> Sure, good point.
[18:37] <barry> xnox: mostly i just want to either trust just the system ones, or the system ones + my self-signed one
[18:38] <aquarius> cjwatson, sure (although if it's not derived from the URL then it might as well be a uuid or something and then it'll never collide). Definitely need the API then to say "what was the download location"? Making the names be meaningful is a little bit nicer for developers, I think.
[18:38] <barry> (where self-signed will either be a "good" one or a "bad" one)
[18:39] <barry> otoh, with d/l service integration these seem more like tests of the d/l service than the system-image code, so i dunno.  (it makes lots of sense when i was testing the built-in downloader)
[18:39] <xnox> barry: how soon is gpg check? e.g. how much is downloaded over https, before gpg check fails? e.g. is it a small json file, or like full image?
[18:39] <aquarius> QUESTION about the whole object paths thing -- I'll only be able to ask about downloads that my app requested, right? I can't just pass a random object path and find out abot it? ;)
[18:40] <aquarius> that sort of thing makes jdstrand come round and punch you ;)
[18:40] <mandel> aquarius, no! :)
[18:40] <fagan> Is it wrong that I read aquarius' comment in his voice :D
[18:40] <aquarius> I didn't know you could limit access to object paths with D-Bus. Cool.
[18:41] <tedg> aquarius, It's with the apparmor confinement we built with dbus.
[18:41] <tedg> aquarius, So you need special dbus :-)
[18:41] <slangasek> yep, it's the New Hotness
[18:41] <aquarius> very cool.
[18:41] <aquarius> something else I have to read about at some point :P
[18:42] <jdstrand> hehe
[18:42] <tedg> aquarius, Open Source, obscurity via too much information ;-)
[18:42] <aquarius> please please please bind the library to QML. That's what everyone's writing apps in ;)
[18:43] <barry> xnox: for the data files, we download everything before doing any sig checks
[18:44] <aquarius> One of the things that needs doing (and resourcing) is adding documentation for the QML API for the downloader to developer.ubuntu.com...
[18:45]  * lool moves to another session real quick
[18:45] <xnox> barry: so Verisign can leak an ssl cert, and I can start endless loop of downloading gigs of updates, all of which fail to verify.
[18:45] <barry> xnox: oh well :)
[18:45] <xnox> barry: that's why i would have thought you (a) want to only ever trust one ssl cert or (b) download metadata file & gpg verify that, before downloading the rest.
[18:46] <barry> xnox: i think those are two different things though.  if the cert is bad, we won't even connect to the https and it will fail early
[18:47] <jdstrand> mandel: is there a cli app to play with?
[18:47] <aquarius> xnox, if you only trust one cert and something goes wrong on the server so a new cert is required, then they'd have to deploy the new cert to all devices... which is a chicken-and-egg problem ;)
[18:48] <jdstrand> should I prefix that with QUESTION? I shall
[18:48] <jdstrand> QUESTION: is there a cli app to play with?
[18:48] <jdstrand> (sorry if this was answered before, I was in another session)
[18:49] <cjwatson> dbus-send? :-)
[18:49]  * aquarius laughs
[18:49] <jdstrand> I guess I could look at the test cases
[18:49] <cjwatson> (You can also toggle it in on the front panel)
[18:49] <cjwatson> Sorry, I'll go and be unhelpful somewhere else :-)
[18:49] <aquarius> I was goig to suggest "echo stuff into /dev/kmem" :)
[18:49] <jdstrand> :)
[18:50] <aquarius> jdstrand, mandel did say that there's a library for this stuff, so having a cli app which talks to that ought to be easy, but I don't know if he's *done* it yet :)
[18:52] <aquarius> need the heuristics, indeed, otherwise my app can say "download http://evil.com/endpoint-which-is-slow-and-never-terminates" and essentially DoS the download service ;)
[18:52] <fugue88> Or self-tuning: add another download, sample total throuput.  Increase?  Yes->add another download.
[18:54] <jdstrand> yeah, I was just thinking about testing out the apparmor policy group for this
[18:56] <aquarius> how does the user pause and resume a download?
[18:56] <aquarius> they have no UI to do that
[18:56] <fugue88> Remove the battery.  ;)
[18:57] <aquarius> tedg says that putting in-progress downloads in the sync menu is planned but isn't on anyone's todo list yet ;)
[18:57] <gatox_> aquarius: from the app that trigger the download probably
[18:59] <fugue88> aquarius: I'll add it to the list (doesn't commit anyone to do it, just make sure we don't drop it).
[18:59] <fagan> I dont thinkk people would mind if you swore mandel :D
[18:59] <fagan> Or is there some rule that I haven't heard about for UDS where cursing isn't allowed
[19:00] <tedg> mandel, We should figure out the sync menu stuff.
[19:00] <tedg> mandel, I think it's a post 13.10 thing though.
[19:00] <mandel> fagan, he, I swear a lot :)
[19:00] <fagan> Yeah I know dude
[19:00] <mandel> tedg, yes, 100& agree
[19:01] <fagan> < shane your long lost intern
[19:01] <mandel> tedg, atm no one is showing the current downloads and is a 'bad' thing
[19:01] <mandel> fagan, I know you ;)
[19:02] <fagan> Ah ive been gone a while so you can never know :P
[19:06] <slangasek> https://plus.google.com/hangouts/_/006bf4a2afd21e219c0b18cbd32a8d53e9dc89bb
[19:06] <slangasek> infinity: https://plus.google.com/hangouts/_/006bf4a2afd21e219c0b18cbd32a8d53e9dc89bb
[19:08] <slangasek> jdstrand: are you joining this session?
[19:08] <xnox> if there is space i'd join =)
[19:09] <slangasek> xnox: there is
[19:09] <jdstrand> ah, missed the url
[19:14] <roadmr> 5~/part
[19:14] <roadmr> oops sorry
[19:33] <xnox> slangasek: and the terminal app is a core-app ;-))))
[19:34] <rickspencer3> o/ sorry
[19:35] <xnox> rickspencer3: hello =) join us https://plus.google.com/hangouts/_/006bf4a2afd21e219c0b18cbd32a8d53e9dc89bb
[19:39] <slangasek> xnox: hmm? what about the terminal?
[19:47] <xnox> slangasek: the train is gone.... =) "infities toggle to install latest image anyway" will be exposed in the UI, inside the terminal app =)
[19:51] <beedub> yes!
[20:00] <beedub> daily-proposed will work after system-image 1.3 ;)
[20:00] <MikeVax> exit
[20:04] <slangasek> infinity: ah, yes - text ui is still ui ;)
[20:06] <infinity> stgraber: If you want to brainstorm random implementation details, I'm happy to babble on Mumble or IRC or whatever over the next couple of weeks.
[20:06] <stgraber> infinity: ok, cool
[20:08] <stgraber> the overall plan is fairly simple, we want one channel per series, one proposed channel for each of those and one customized channel for each of those too. Then we want three fake channels, daily, stable and lts (not sure about that last one) which are copies of their series equivalent with only the version tarball re-generated (which contains the channel name and version)
[20:08] <stgraber> now implementing that mess to avoid duplicate files on the filesystem, generate the right deltas, have a way to release good images, ... is where the fun is at
[20:17] <xnox> stgraber: i don't think lts makes sense on the phone at all, at the end of 2 years, one throws away the phone & gets a new contract and "syncs" everything from the cloud.
[20:17] <xnox> stgraber: and I highly daubt a two year old phone will be able to run new OS. It would be awesome if it did, but in practice this has not yet been the case with any mobile OS.
[20:18] <xnox> or any mobile OEM (to exclude cyanogenmod)
[20:21] <xnox> ... well if we keep hw layer same and upgrade ubuntu rootfs, we'll know in 16.04 =)
[20:22] <cjwatson> I thought we were trying to do better than previous mobile OSes :-)
[20:23] <xnox> =)))))
[20:25] <stgraber> once I have a way of aliasing channels, it'll be trivial to add more, so I don't think having an lts alias would hurt