mozilla :: #mobile

15 May 2017
14:31nine_millihey i think i may have found a bug
14:31nine_millihiding then showing an iframe results in all selects and inputs no longer selectable
14:34nine_millianyone know if there's a workaround to this issue?
14:36rnewmannine_milli: did you file a bug?
14:43nine_millinot yet
14:43nine_milliwanted to verify if it is one
14:44nine_milliok removing and readding works whereas hiding showing doesnt
16:35mkaplyI thought the table UI had a home button?
16:35mkaplytablet UI
16:56kbrosnanmkaply: only in large tablets
16:57* mkaply was hoping to reuse that code to put a homebutton somewhere on the phone
17:47mkaplyIs this build
17:48mkaplyan old aurora or a current nightly?
17:55ahuntI think thats up-to-date aurora
17:55ahunt(I dont believe Aurora has been remove yet?)
17:56ahuntAre we still going to be providing the mozilla update service when aurora disappears (I know the plan is to provide nightly via play, but does that imply the mozilla update channel is going to go away?)
19:21jhugmanrnewman I think I made a mistake in my notes from Friday I wrote down SyncStateMachine to add an optimistic mode; I think I meant to write an FxALoginStateMachine. Could you confirm that?
19:27WG9s_jhugman: Is that the past Pluperfect tense?
19:29jhugmanWG9s_ "pluz"? It's the en_GB spelling of "pls"
19:37jhugmangah, no. We spent most of the time talking about MetaGlobals etc, though I do have written down a married state. No, I think I was right the first time; I need to add an optimistic mode to sync state machine.
19:38WG9s_jhugman: but this is an old joke about us form Boston about a tourist getting in a cab in Boston and asking where he can get Schrod (a baby whitefish usually cod). The Cab driver replies, "Boy I have herd that questions a million times, but never before in the past plu-perfect"!
19:40WG9s_THe only joke I know of that would combine the knowlege of a Boston Cabbie with an English grammar teacher so you get the point.
19:43jhugmanoh ha. I've not been to Boston before.
19:48nalexanderjhugman: rnewman is mostly PTO the next 2 weeks. I can probably help with your state machine question. Context?
19:49rnewmanjhugman: short answer: both!
19:49rnewmannalexander: syncing on push received, but only if our FxA is good and we can go straight through HasMetaGlobal and HasCryptoKeys to Ready without blanking collections or syncing from scratch.
19:50* nalexander backs out
19:50rnewmanthat is: we want the cheapest possible sync, and otherwise do nothing
19:50rnewmanbasically we want to pass a set of excluded (or included) states to toReady
19:50rnewmanand we will fail to transition through the state machine if we can't follow the good path
19:51rnewmanwe want to do the same thing for FxA, but I suspect it's all cheap enough or will fail at the Sync hurdle
19:53jhugmanI'll experiment with not doing it with adding an optimistic mode for FxALoginwe can probably manage just by checking .actionNeeded
19:59jhugmanerg. no NeedsFreshMetaGlobalState
19:59jhugmanerg. no NeedsFreshMetaGlobal state
20:41snorpalso if anyone wants to see jchen cry this is the time
20:41snorpor me
20:43jcheni was not emotionally prepared for that
20:44snorpwebrtc java stuff...kinda a mess
20:47jesupsnorp: jchen: we could replace all the upstream code (like we did for audio input), and then maintain it all ourselves. No small effort (and our webrtc team aren't Java/Android experts; the integration with the upstream code was largely done by gcp). Also, before crying too much, you should look at the tip of code at that we're trying to catch up to with imports of newer versions
20:47snorpjesup: is the upstream code better
20:47jesupWe can also rewrite stuff and push upsteam to accept it - we're trying to do that more; but it can be hard
20:47snorpjesup: I think we can work around the specific problem I'm having here
20:48snorpbut hand-coded JNI is fuuuuuugly
20:48snorpand error prone
20:48jesupsnorp: Haven't looked
20:48jesupand honestly don't really know what to look for, not being a java/android person
20:49gcpI think I wrote this.
20:49gcpUpstream imported a JSON library on both sides and passed everything over a string.
20:49gcpIIRC it's been years
20:49snorpgcp: wait what?!
20:50gcpChrome does the same by compiling in a full JSON library, twice.
20:51snorpthat's the problem
20:51gcpdid not write that
20:51gcpI think that's our own code tho
20:51snorpyeah I think that is
20:52snorpwe need that permission request to go through something else
20:53snorpideally, I think there would be a gecko permission for camera
20:54snorpand then GeckoView or Fennec would handle that and request the android perm
20:58jesupgcp: snorp: they removed that file from upstream in 49 (last import); we restored it (we had mods to it). Not sure who did it; pkerr likely or maybe gcp.
20:58snorpjesup: interesting
20:58jesupdetails are hidden in
20:58gcppkerr did that, I just tested the outcome
20:59snorpso we basically own that file?
20:59snorpthat's good I guess
20:59jesupwe do now
20:59gcpclear sign we own the file :P
20:59jesupwhich means we don't have to worry about upstream for mods - but may need to worry about it failing due to upstream changes
21:00gcpAndroid 2.2/3 support
21:00jesupvideo still works in the v57 import dminor and I are working on - close to reviews; dminor did the android part and it wasn't too bad apparently
21:02jesupBug 1341285 -
21:02firebot NEW, Update WebRTC code to stable branch 57
21:03jesupand that file is still in the v57 setup - we didn't remove it.
23:52Caspy7anyone know if Desktop firefox can still cast to Roku?
16 May 2017
No messages
Last message: 12 days and 1 hour ago