mozilla :: #mobile

15 Mar 2017
02:13walkingiceI tried nightly (from yesterday) and the loading bar works normally. (I opened a google page)
06:40sebastian^--- jack is history again.
06:41sebastianUpdating the SDK might be a little bit easier with that.
07:06Gautam-MJust out of curiosity, I see this message on many documentation pages - 'The xyz object is only available to privileged code running on abc'. What does this means?
07:44ahuntGautam-M: do you have a link? I think the privileged code is just to make clear that a random webpage cant access xyz, but browser (and possibly addon) code can
08:14Gautam-Mahunt: Yeah, got it. :) I shouldn't have used 'abc' there, I meant browser/add-on only. Thx.
10:41AutomatedTesterioanachiorean: how are you feeling today?
11:16ioanachioreanAutomatedTester, alive - I do not have muscular fever - I feel a bit the feets only when taking the stairs
11:16ioanachioreanbetter than expected to be honest
11:16ioanachioreanthanks for asking
11:17AutomatedTesterioanachiorean: Tomorrow might be worse!
11:18ioanachioreanthat is my feeling too to be honest
11:18AutomatedTesterioanachiorean: but youre currently 3rd in the strava leaderboard with that distance \o/
12:38ioanachioreanwow AutomatedTester that is cool! Brone it is cool for me
12:38ioanachioreanI guess davehunt is teh 1st
13:14AutomatedTesterioanachiorean: nah, he is #2
13:14ioanachioreanoh no
13:15sebastiando you still need a last? i could be the last one in your strava leaderboard
13:15snorpsebastian: yesssss
13:16snorpsebastian: we should immediately use java 8 crap once that goes in
13:16sebastiansnorp: Yeah! Absolutely. I really hope we can finally update the SDK.
14:14snorpstandups: still debugging SurfaceTexture test failures, probably need to hook up attach/detach to allow snapshotting of WebGL and videos
14:14standupsOk, submitted #43757 for
14:27franksebastian: did you see the new ping format doc?
14:30sebastianfrank: Um, no! :)
14:31frankthat ping solves the issue of "which percentage of users are using setting X"?
14:31frankand I've added the settings change to the events list
14:31sebastianah, perfect!
14:32franksebastian: feel free to comment on there or our initial sheet
14:33frankand if you have more questions we should get together again
14:33* frank hopes his internet doesn't crap out :(
14:33sebastianfrank: But we still send a core ping, right? Some data is now duplicated .. is that to make the processing simpler?
14:34franksebastian: yes, to make the analysis easier
14:34franksebastian: do we have the core ping for focus?
14:34frankI thought you mentioned it was a "maybe"
14:36sebastianOh, did I? I'm implementing both ping types currently - although I'm not sure whether this is strictly required? I guess I'm missing knowledge about the analyzing part for that.
14:37frankoh man, that's great!
14:37frankno, it shouldn't be an issue
14:37frankhaving this be a separate ping makes a lot of sense, especially if we implement some sort of throttling for data reasons
14:37frankand to do analysis, we would need data in the ping, so it's OK if it's duplicated
14:38frank(( it is very difficult to correlate different pings together, e.g. core and event ping ))
14:40sebastianfrank: Afaik for Fennec both ping types end up in two tables that can be queried via re:dash (and joined with the clientid) - Is that something that is/was hard to do?
14:40frankno they definitely can be
14:40frankbut the join is prohibitively expensive
14:40sebastianoh, i see
14:40frankbecause really we would need to get the *latest* core ping per event ping
14:47sebastianfrank: Is there any server we can send test pings to? Mh, I remember using this locally in the past:
14:49njparkahunt: so is the syntax for error page about2:error:ERROR_IO for example?
14:50sebastiannjpark: It should be something like error:-2
14:50njparkoh, no about2: anymore
14:51sebastiannjpark: Yeah, you can see the errors here:
14:51sebastiannjpark: And then you need to use the constant value, that you can see here:
14:51njparksebastian: ahunt: awesome, I see it, thanks!
14:51njparki guess we can close this bug then?
14:51sebastianyeah :)
14:54franksebastian: that seems right to me. Let me ask around to make sure
15:05sebastianfrank: For release channels - do we use different endpoints? Maybe not super important right now with only ~20 people with test builds.
15:06franksebastian: nope, same endpoint
15:06franksebastian: look at the core ping submission here:
15:06frankspecification has the appUpdateChannel
15:06frank /submit/telemetry/docId/docType/appName/appVersion/appUpdateChannel/appBuildID
15:07sebastianfrank: ah! bonus question: do we already know what we pick for appName etc. and at what point could we send data to the actual endpoint? :)
15:08frankhmm. Nope I do not!
15:08frankprobably focus?
15:08sebastianand iOS / Android is filtered based on os in the ping?
15:09frankand to the latter question, not yet. I need to get the schema right first
15:09franksebastian: yup, that's how we differentiate the os's
15:10sebastianfrank: (and maybe that's a question for barbara actually) do we need to distinguish between Focus and Klar builds?
15:10frankyeah I'm not sure about that
15:10frankbut I feel like we should in some way, right?
15:11frankbtw we need to come up with a docType as well for the new ping
15:11sebastianyeah, without making evaluating events hard though - for most questions the build doesn't matter
15:13franksebastian: hm. Yeah I'm not sure if the analysis will need it. Best to ask Barbara
15:15barbarafrank: sebastian; it would make sense to differentiate between the two builds, helps us understand what market is most successful -- would this cause too many different meta information for the events?
15:16frankI don't think so, we just tack it on the ping
15:29mkaply bnicholson (thought you might know): Is migrating list.txt files for iOS something that is done on a regular basis? Or only done as needed?
15:29mkaplyI know Focus is using the new JSON file
15:39flodas far as I remember it was done "as needed"
15:39flodand we probably didn't do that in a while
15:45sebastiandelphine: flod: Do we have a translator or someone else from the community who can help us validate RTL support in Focus? -> - Someone who actually speaks one of those languages and knows what's going on :)
15:47sebastiannjpark: For the screenshots of the edit mode: Could you just enter 'mozilla'? It should autocomplete to in the URL bar and just show "Search for mozilla" below.
15:47sebastiannjpark: That would be helpful to see a difference (url vs. actually entered text) :)
15:48njparksebastian: sure will do, I was going to create separate UI tests for autocompletion anyways
15:48njparkhmm, I guess I need google play app to open the market:// link, maybe I need to run it on actual device
15:50sebastiannjpark: Mh, if there's none on the emulator, maybe you could add a custom URL handler for "market://" to the manifest of your test package. A bit hacky but could work. ;)
15:51njparksebastian: hmm, could you point me to a sample? I've never done that before
15:53sebastiannjpark: Like we, as browser, respond to http://, your test package might be able to response to market:// ->
15:54sebastianMaybe Android prevents that though.
16:06flodsebastian: delphine has the right contacts in our community for helping with RL
16:07delphinesebastian, flod: yes I can definitely help with that
16:11sebastiandelphine: Awesome!
16:12delphinesebastian: I'll reach out to some folks today :)
16:12delphineand also will take a better look
16:13delphinenjpark: hey! I was wondering if we could get a new round of localized screenshots some time soon, since we've had new strings landing :D
16:13delphine(for focus on android)
16:14sebastiandelphine: njpark: I guess I should do another import today :)
16:16delphinesebastian: oki doki!
16:16mkaplyflod: I just want to make sure no one attempts it soon. I'll need to redo my patches for iOS to use the JSON files
16:17flodit should be easier to check the history, I haven't done that in a while
16:17justindarcsebastian: frank: catching up on the scrollback now.. are we still going to try and meet later this week?
16:18flodmkaply: 9 months ago, as long bnicholson is aware, nobody is likely to touch them ;-)
16:18mkaplyGood :)
16:18flod(7, not 9)
16:18frankjustindarc: I'm out tomorrow and Friday :(
16:18frankI'm putting together an email now with HTTP request specs
16:20justindarcfrank: ahh ok.. thanks for putting that together.. i'm working on getting a mock HTTP server up and running for unit testing the iOS library
16:22frankjustindarc: nice. Let me know if you have any other questions
16:22frankjustindarc: also please feel free to comment on the proposed ping format
16:23justindarcfrank: will do, thanks!
16:24ahuntHuh, buddybuild is failing on a search engine resource (with a different locale being the culprit each time):
16:25ahuntThe original commit seemed to be fine (i.e. buddybuild didnt complain then), but now that its on master, my PR is failing.
16:30sebastianahunt: For some reason it goes away after clicking "Rebuild without cache"
16:33ahuntIm getting the same locally now too :)
16:36sebastianahunt: I think I saw it once too and it disappeared after a clean too =)
16:36sebastianSomething is flaky
16:36sebastianfeels almost like building fennec ;)
16:38* ahunt tries to escape
16:39ahuntAFAICS these are just normal assets, so thats actually just gradle being flaky?
16:39sebastianI hope you do not have a new MacBook *escape-joke*
16:40sebastianahunt: Yeah, there shouldn't be anything special
16:50sebastianahunt: Oh, the toolbar is broken on master when scrolling?
16:54ahuntahunt: it seems to work for me, both on a custom build, and with buddybuilds last beta
16:54sebastianit only happens for VIEW intents I send from the command line. let's see if this happens with third-party apps too
16:56sebastianargh. yes it is.
16:56sebastianI'll file
17:05ahuntOh, I see it too - only if I completely kill focus first
17:08sebastiantoo much Android anomaly today
17:09njparkdelphine: will do, I'm working on creating another test just for the error msgs though
17:22rfeeleygrisha: whoah, 41% of fxa avatar changes come from android
17:36grisharfeeley: interesting, hopefully we're correctly collecting this data.
17:37grisharfeeley: speaking of avatars, do you know what our rationale was for including them? it seems like a very strange thing to have in a browser account (unless one might try to use it as a more general identity)
17:37delphinenjpark: sure! thanks a bunch!
17:38njparkdelphine: yeah probably by eod I'll have a new set uploaded, I'll try it on real device this time
17:38rfeeleygrisha: thats the longer term goal, get it in there for profiles. but profiles are slow to come.
17:43grishasebastian: can we open a tab in the background, without having it load until accessed?
17:46sebastiangrisha: Yeah, that should be possible. I don't know how from the top of my head though. Maybe start with the load flags in
17:46grishasebastian: ok, great, ty
17:47sebastiangrisha: LOADURL_DELAY_LOAD and/or LOADURL_BACKGROUND sound helpful
17:51sebastiannjpark: You should be able to select and use different search engines now. Oh and when running a focusWebkitDebug build you won't need to change the setting anymore to screenshot (***Debug builds will allow it)
17:53njparkSebastian: okay thanks!
18:03grishasebastian: how difficult might it be to add a tab to tabs tray when browser isn't running? i guess that's part of the session restore process?
18:04grishaa hack i can think of is persisting information about a tab in some sharedpref, and then opening it after launch
18:05sebastiangrisha: This sounds like what tab queue does? As long as you do not want to do actual loading.
18:05grishaahh right, i forgot about tab queue
18:13nalexandergrisha: rfeeley: I did a lot of the work to add avatars into Fennec with vivek (a longtime, now departed -- boo -- contributor)
18:13nalexandergrisha: rfeeley: the motivation was always to make Fennec "yours".
18:13nalexandergrisha: rfeeley: to increase people's sense of ownership of the product and experience.
18:15sebastianit definitely does that for me.
18:16rfeeleynalexander grisha sebastian another interesting graph. users manipulate their uploaded images more often than not.
18:17grishanalexander: i have to say, after setting an avatar a last week, the browser does "feel" a little different to me. it's subtle, but effective. especially in light of that avatar appearing on three instances of it
18:18nalexandergrisha: yeah, I've always wanted to put the avatar in the Fennec Settings menu, but others (a.ntlam?) were generally against.
18:20grishanalexander: perhaps w/ photon
18:27njparkSebastian: hmm i found a bug when running on device- back button in settings exits the app when i switched to it from app list view- i'll log the str
18:33sebastiannjpark: Yeah, please file! :)
20:25antlamnalexander grisha: in Settings? I remember mocking something up for a contributor actually. I liked it but I don't remember what came of the bug :\
20:26grishaantlam: i think i saw a bug you're talking about a few weeks ago.. (also can't quite remember)
20:27grishaantlam: will there be changes to how fennec's settings menu/popup look like in photon?
20:32nalexandergrisha: antlam: are there Photon changes for Fennec at all? I expected it to be Desktop only
21:00ahuntsebastian: just in case there are still string import issues (Im not sure if there are system specific issues even with the latest fixes), Ive pushed my own local copy to a branch:
21:01ahunt(I could probably add an extra layer of hacks to remove the xmlns part, which is completely unnecessary but a side effect of the xml library)
21:39njparkdelphine: have to catch train now, but will upload the screenshots once I get home. as sebastian says, the error msgs are not localized so I won't include them
21:39delphinenjpark: sounds good, thank you :)
22:38grishanalexander: if you have a sec:
22:38firebotBug 1346438 ASSIGNED, Use 'X-If-Unmodified-Since' header when uploading meta/global and crypto/keys to avoid race conditio
22:39nalexandergrisha: reading
22:39nalexandergrisha: if you do as you say -- having request and delegate both contribute XIUS -- you have violated "singel responsibility".
22:40nalexandergrisha: and we set the pattern (delegate owns XIUS) long ago.
22:40nalexandergrisha: I think we'd do it differently now, just having the request own all that stuff up front. But we shouldn't switch horses midstream.
22:41nalexandergrisha: if you feel strongly, I'll take it... but you'd need to document which took precedence, and document what "absent" value is, etc, etc.
23:28njparkdelphine: hmm, is fy-NL a correct locale? it shows everything in English (and seems to be down for me now)
23:31delphinenjpark: I'm wondering, it's maybe because Android doesn't support it. We do with Firefox for Android but if there's no language switcher on Focus Android, that may be why? What do you think @sebastian?
23:31delphinenjpark: because we do have it on Android for Firefox:
23:47delphine(nevermind we noticed in the meantime)
23:47RaticideYo! So I've been using Firefox on android for ages, but since the last update it no longer loads pages. Just a white screen. Is there a fix in the works?
23:47RaticideThere's a massive load of 1 star reviews on the store reporting the same issue.
23:48RaticideThe beta version is still 4.3 and works great. The stable version is 4.4 and is completely broken.
16 Mar 2017
No messages
Last message: 13 days and 8 hours ago