mozilla :: #mobile

17 Apr 2017
01:51justdaveso I'm flying tomorrow, just tried to check in, and the e-boarding pass they sent me won't display the barcode in Firefox (either mobile or desktop)
01:51justdavewould normally file a webcompat but the URL exposes my ticket and I'm not actually on the flight yet.
01:52justdaveor Delta's system is just broken, doesn't work in Safari or Chrome either
01:56justdaveoooh, URL has a trailing slash on it, K-9 didn't include the slash in the link. And Delta's website tries to be smart and shows it anyway, but doesn't redirect and the lack of a slash breaks the relatvie link to the barcode
02:01justdaveand that was fixed in k-9 26 days ago, they haven't pushed it to the play store yet
02:02justdaveer, actually they did, and it's still broken :(
02:04justdaveoh, release version is on a branch :|
14:14wlach|afksnorp (or anyone who knows): I'm testing mozregression on android for the first time in a while. is there any way to disable the android updater on nightly/inbound builds? looking for something in the source tree but not finding it so far
14:15wlachoh huh, found it "app.update.enabled". but mozregression is already setting that
14:16snorpwlach: you cannot disable the update check
14:16snorpwlach: you can disable automatic download of the update through app settings
14:16snorpbut can't set that through mozregression IIRC
14:17wlachsnorp: ?
14:17wlachsnorp: trying to set a sane default
14:17snorpwlach: I don't know if we use that
14:18wlachsnorp: it appears we do
14:18snorpwlach: excellent
14:25wlachsnorp: in fact setting app.update.enabled to false is enough, mozregression wasn't setting it for fennec
14:25snorpwlach: interesting
14:25wlachmozregression really needs an actual maintainer :/
14:26wlachas opposed to a fake maintainer, like wlach
14:26snorpwlach: wasn't there a contributor working on it a lot
14:27wlachyeah, parkouss-- he found another job :)
14:27wlachpaul yang from taipei has been helping out a bit, but it's one of many things he's working on
16:01nalexandermcomella: re: Android Account and signing keys -- sharing the AccountManager managing code is not usual, and requires the same key _and_ the same sharedUserId (I think).
16:02nalexandermcomella: trying to abstract this Account stuff behind an authenticator has made very little sense historically, since you have to "hand over" the Sync key to any consumer for them to do things.
16:02nalexandermcomella: but with oauth tokens and non-Sync FxA-connected services, that's no longer the case. However... it's a huge pain to actually ship a separate authenticator APK; the only folks who do "own the phone" -- Google, Samsung, Tizen, etc.
16:19snorpjchen: so so don't seem to have any cases where the java code actually does anything in disposeNative()
16:19snorpjchen: if I want to destroy from that end what do I need to do?
16:21jchensnorp: destroy from java?
16:21snorpjchen: all of the destroyNative() impls I've seen just throw an exception
16:21snorpjchen: since you're not expected to call it
16:21snorpjchen: but if I want my java instance to destroy the native one there how do I do that
16:21snorpjchen: I had misunderstood your earlier review comment re: the surface texture listener
16:22jchensnorp: usually you make disposeNative native and call that
16:22jchensnorp: e.g.
16:23snorpjchen: ah, and that's already implemented in the base native impl
16:23snorpso it just does the right thing
16:23snorp(unref, delete, etc)
16:24snorpok that was definitely not obvious
16:24* snorp will add a comment
16:44njparksebastian, ahunt: who would be a good person to ask if I want to learn more / discuss about code coverage / unit tests for fennec android?
16:48nalexandernjpark: mcomella has historically been interested. I can help a little with build / Task Cluster stuff.
16:48njparknalexander: aha, thanks! so we already collect code coverage data?
16:49mcomellanjpark: As far as I know, we don't
16:49mcomella(my focus area was more getting unit tests working in m-c & various linters)
16:49mcomella(I didn't really look into code coverage options)
16:50njparkmcomella: aha :), i was looking into jacoco stuff, awesome, would it be okay if I ask you regarding unittests/lints from time to time? Also, is there a wiki where I can get a overview of those?
16:51mcomellanjpark: Of course you can ask (though I'm not sure I'll have relevant answers any more)!
16:51mcomellanjpark: may also not be up to date but that was the state of the world when I was working on firefox
16:51mcomellanjpark: The quick reference has the things common to Android dev's workflows, while the rest of the page has details on more gecko-related stuff (e.g. xpcshell tests)
16:52njparkmcomella: one of the PI team's okr is 'enforce unitest/lint/code coverage' but I want to just start off with learning about the current state and find areas where I can provide help, great! I'll read up on those and get back to you, thanks!
16:54nalexandernjpark: sylvestre was also very interested in automated testing, etc.
16:55nalexandernjpark: and he was running static analyzers against Fennec at one point. Not sure code coverage was part of that.
16:56njparknalexander: thanks for the info! yeah it seems like there are quite a few tests running for fennec, and we probably need to decide where we collect the coverage data... will look into that as well
16:58nalexandernjpark: in general, it will be hard to get meaningful code coverage, because of Robocop and the various test runners (mochitest-{browser,chrome}) that are invoked in very different ways.
16:58nalexandernjpark: code coverage of units tests should be well handled by Android Gradle -- or not worth doing :)
16:59njparknalexander: we do have some junit4 tests right? yes that's what kbrosnan suggested as well
16:59nalexandernjpark: yes, those ones should be easy to code coverage.
17:00mcomellanjpark: btw, here's an opinionated overview of how we would use the various test frameworks:
17:00njparknalexander: yes i think it's best if I start small.. hopefully the gradle tool is easy to experiment
17:01njparkmcomella: thanks! this is great help, so tests like autophone mainly belongs to perf testing i suppose?
17:02bcnjpark: not really. they also supported a subset of the unit tests.
17:02njparkbc: i see, i should really sit down and read through all the tests, thanks!
17:05njparkbc: ah, but collecting coverage data might be difficult for autophone since it runs separately on device, i see
17:09bcprobably the case, yes.
17:26nalexandermcomella: for what it's worth, I concur with your opinions :)
17:27nalexandermcomella: I would add some notes about mochitest-browser, which gbrown has set up to "steal" Robocop JS tests.
17:27njparkrobocop tests sounds pretty unloved
17:29nalexandernjpark: actively despised, I woudl say.
17:35sebastiannjpark: I use jacoco for the telemetry library (code coverage of the unit tests). It's pretty nice so far!
17:50mcomellaUnfortunately, I don't know about mochitest-browser
17:50mcomellaEnough to add notes
17:54njparksebastian: i might bug you later about it after reading more about that, maybe we could put it in focus?
17:55sebastiannjpark: Yeah, I'm on PTO today. But I'll be back tomorrow. :)
17:55njparksebastian: i am sick today :)
17:55njparkso yeah tmr sounds good
17:58sebastianOh, an android police article!
18:04nalexandersebastian: did you just flip Jacoco on in Gardle?
18:44sebastiannalexander: Yep. And upload the reports to to get it integrated into GitHub.
18:44nalexandersebastian: aye.
19:09RyanVMcustom tabs are still nightly-only, right?
19:11JanHRyanVM: yup (
19:13RyanVMthanks :)
19:14RyanVMOOC, any ETA on when they'll ride the trains?
19:16JanHstandalone mode webapps (which are similar in some regards) are planned for 55 I think, and according to the same is apparently true for custom tabs proper
19:17JanHbut for absolute confirmation you'd need to ask somebody from the Taipei team
19:59nine_millianyone know what i need to implement to get select boxes working in geckoview?
20:00jchenbug 1322581, which is ready to land actually
20:00firebot ASSIGNED, [geckoview] Move prompt-related delegate function from ChromeDelegate into PromptDelegate
20:46nine_millihey jchen, so in a few days i should be able to do a mercurial update to get it?
20:46nine_millior should i download a patchfile?
20:48jchennine_milli: yeah. you can apply the patches too if you update them to the current tree first.
20:48nine_milliok thx
20:55mcomellagrisha: I received the returned credentials from the web flow (uid, keyFetchToken, sessionToken, unwrapBKey) and I'm trying that these credentials are valid - do you have suggestions on how to do that? I've been trying to call methods on FxAccountClient20 but I'm getting 404 - I wonder if it's because the dev server I've been using doesn't support methods
20:55mcomellalike `deviceList`and `keys` (on FxAccountClient20)
20:56mcomellaI've been using the url
20:56nalexandermcomella: the dev and production servers do not share a store. So you'll need to ensure the entire server endpoint constellation agrees.
20:56nalexandermcomella: that is, fxa-{auth,content,profile}-server, and token server (which ensures you get the right Sync endpoint).
20:58mcomellanalexander: I don't actually know about these separate servers (auth,content,profile) - are there docs where I can read more?
20:58nalexandermcomella: yes; you don't need to understand much, but they do need to line up.
20:59mcomella(so I'm guessing I've been trying to access the auth server, rather than one of the others)
20:59nalexandermcomella: let me find links, one second. For now, look in the Firefox iOS code for "configuration" -- that's most ofthem.
20:59mcomellaAh, yes
20:59nalexandermcomella: well, you'd need an auth server and a content-server at least. The content-server serves about:accounts; the auth server backs it.
21:01nalexandermcomella: -- not sure where it's relaly documented; reverse engineer iOS (which is simpler than Android)
21:02* nalexander grooves to the nostalgia of the Super Castlevania soundtrack
21:06grishathanks for answering, nalexander .
23:00ahuntIt looks like MSs arrow launcher also has its own (OS-independent) locale selector - I wonder how many other apps do that
18 Apr 2017
No messages
Last message: 12 days and 8 hours ago