mozilla :: #mobile

16 May 2017
01:06concernedanyone here?
12:32shatur-Mhey!! When I toogle "privacy.item.cache" or any other "privacy.item" in about:config, it does not reflect in "Clear private data" in settings. Are these two different things??
12:34shatur-Min contrast to when done in desktop firefox ,i.e., toogling "privacy.cpd.*", it does reflects in settings
13:10snorpshatur-M: probably different, yes
13:44rnewmanshatur-M: Android clear private data is different. most of the gecko prefs don't apply.
14:51snorpjchen: ping?
14:51jchensnorp: pong
14:51snorpjchen: GVA uses BuildConfig.DEBUG, but that doesn't seem to exist?
14:51snorpjchen: even though it does in fact build
14:51snorpwtf kind of black magic is that
14:51snorpI guess I'm looking at the wrong file?
14:51jchensnorp: it's autogenerated based on gradle config
14:51snorpoh, geez
14:52snorpok totally separate from the gecko BuildConfig of the same name
14:53jchensnorp: it's kind of hacky but we use "" when building with mach, and autogenerated BuildConfig when building with gradle
14:53jchensnorp: the two are kept in sync manually iirc
14:53snorpthe gradle one has nothing from the other one
14:54snorpoh nm
14:54snorpapparently it gets processed further on
14:58jchensnorp: oh i guess that's just for geckoview, for geckoview_example i think there's only one BuildConfig (that's autogenerated) because geckoview_example is gradle-only
14:58snorpjchen: yeah, weird
15:48snorp<bkelly> does fennec not support --run-until-failure when doing ./mach mochitest some/test/path?
15:48snorpgbrown: ^^^^
15:48bkellyoh... I posted in the wrong channel... woops
15:49bkellymobile... e10s... whats the difference?
15:49gbrownsnorp: I&#39;m not sure. I think it should.
15:49bkellygbrown: the harness kept failing for me when I added that option to it
15:49bkellywithout running any tests
15:49gbrownI&#39;ll have a quick look
15:49bkellyI guess I can file a bug
15:50gbrownsure. cc me.
15:52bkellygbrown: do you know what the timeout value for mochitests on android debug is? 60 seconds?
15:53gbrownbkelly: sorry, I&#39;m not sure.
15:53bkellyok, thanks
15:56gbrownbkelly: it might be much longer - 5 or 10 minutes...but I may be thinking of the adb timeout or something like that
15:58gbrownbkelly: --run-until-failure worked for me. my command line was
15:58gbrown./mach mochitest testing/mochitest/tests/Harness_sanity --run-until-failure
15:59gbrownand I am on ubuntu 16, and using the emulator
16:01bkellygbrown: hmm, I&#39;ll file a bug with my error messages later today
16:04bkellygbrown: do you know what device we use in automation for &quot;Android 4.3 API15+ debug&quot;?
16:06snorpit&#39;s an emulator
16:06bkellysnorp: I take it thats ridiculously slow?
16:06snorpbkelly: you can&#39;t even imagine
16:07snorpbkelly: actually you can if you run it locally :)
16:07snorp./mach android-emulator
16:07bkellysnorp: how do i run mochitests against it?
16:07snorpbkelly: just like any other device
16:07snorpthe mach command above starts the emulator
16:07snorpand it appears as a device
16:07bkellysnorp: so unplug my device, start the emulator, and then do package/install/mochitest?
16:08snorpbkelly: or just mochitest, yeah
16:08snorpwell you need to install
16:08bkellyoh, mochitest does the package/install?
16:08snorpbut you don&#39;t need to package again
16:08snorpunless you change code
16:08snorp./mach android-emulator && ./mach install && ./mach mochitest
16:08snorpor so
16:11gbrownbkelly: &#39;mach android-emulator&#39; will give you an environment that&#39;s very similar to &quot;Android 4.3 API15+ debug&quot;, but the emulator speed is very much affected by the host
16:12gbrownyour laptop will likely run much faster than the aws hosts we use for the treeherder tests
16:12jchengbrown: btw aarch64 support is almost ready, so we should start looking at adding aarch64 builds on builders
16:13jchengbrown: do you think you help with that?
16:13jchen* you can help with that?
16:14bkellygbrown: you mean my 16 core desktop machine will run faster :-)
16:14gbrownjchen: I haven&#39;t done much with builds...happy to help get associated tests running though.
16:15gbrownbkelly: heh, yeah. ++faster
16:17jchengbrown: ah ok. do you know who usually works on builds?
16:19gbrownjchen:;s been such a long time since we&#39;ve made changes...nalexander did the gradle build work, I think
16:20gbrownmaybe a releng bug with kmoir and me cc&#39;d would be the place to start
16:20nine_millihey everyone, i have a custom protocol blah://, im getting blah://index.html but everything on the page becomes like blah://index.html/jquery.js
16:20nine_millihow can i make it sp a domain isnt requried?
16:24nine_millithe goal is to have it where it become stuff on the page becomes blah://jquery.js etc
16:45snorpnine_milli: not sure it&#39;s possible?
16:46snorpisn&#39;t host a requirement of valid uri?
16:46snorpnine_milli: also I am increasingly curious about what you are up to
16:47snorpanyway, recommend putting some placeholder host in there like blah://blah/jquery.js
16:47* snorp shrugs
17:01ahuntFor custom tabs, Ive been wondering whether we need to sanitise Bundles we grab from Intent extras, since those could have the exact same issue that we had with Intents crashing when the extras bundle is unparceled.
17:01ahuntIve managed to write a demo app that crashes Fennec by doing exactly that, but do we need to worry about real apps doing that?
17:02ahunt(Bug 1365167 )
17:02firebot NEW, Ensure Intent Bundle access is safe
17:08mattwwalkingice: ping
17:08walkingice@mattw pong
17:09mattwwalkingice: hi - I was wondering if you could try to help me more clearly understand your first comment in
17:09firebotBug 1331742 REOPENED, Add support for browserAction.onClicked
17:10mattwI&#39;m not really sure what should be changed
17:12walkingiceI see. I think the naming to be &#39;id&#39; might be better than &#39;position&#39;
17:13walkingiceMenu has several MenuItem, to find a MenuItm by id or find a MenuItem by position is different thing
17:14mattwwalkingice: gotcha, I follow now
17:14mattwthanks, I&#39;ll make that update
17:14walkingicemattw: thanks :D
17:26nalexanderjchen: gbrown: re: adding aarch64 builds, it&#39;s all in tree these days so you can cult the android-api-15/android-x86 definitions in Task Cluster to add jobs, etc. I don&#39;t have time to own this but I can give directions.
17:27nalexanderjchen: gbrown: is a good example to start with.
17:27firebotBug 1291370 FIXED, [geckoview] Add Fennec-and-GeckoView Task Cluster job building using --with-gradle
18:43snorpwoo jchen will be a taskcluster expert
18:44snorpsurely there isn&#39;t so much crap to copy?
18:44jchensnorp: is-such-a-thing-even-possible.gif
18:44snorpI mean we just want to change the mozconfig
18:44snorplike ffs why is there so much stuff to change
19:45Caspy7someone here is seeing black on all screens
19:45Caspy7with no addons
19:45Caspy7perhaps someone can chime in
19:48SeburoClear cache?
19:50AaronMTprobably a bad rom
19:50JanHsnorp: out of curiosity, what&#39;s the idea you have in mind for bug 1346413?
19:50firebot NEW, Going back from a CustomTabsActivity doesn&#39;t trigger application-background notification
19:51snorpJanH: it&#39;ll be tricky
19:51snorpJanH: but basically send background/foreground in response to 0 or more than zero active browsers
19:51JanHsnorp: but then who sets the browser to active or not?
19:52JanHsnorp: GeckoView aside, we could do something like that even today - just send background/foreground based on GeckoApp&#39;s onPause/onResume
19:53snorpJanH: yeah
19:53JanHsnorp: but the complication is that a) switching between different GeckoApps shouldn&#39;t trigger background which is immediately followed by another foreground
19:53snorpJanH: yeah, could debounce
19:54JanHb) so far we also tried to avoid background/foreground events when showing other activities within our app, e.g. the settings
19:55JanHre debouncing, I&#39;m not sure if our current implementation actually works as expected, but wasn&#39;t the original idea that we want to things synchronously within onPause, so Android doesn&#39;t kill us?
19:55JanH*to do things
20:06snorpJanH: that&#39;s true
20:09JanHsnorp: the short term solution would be to omit the finishing() check for custom tabs, since it doesn&#39;t make any sense there
20:10JanHsnorp: what is the proposed timescale for geckoview based custom tabs/web apps?
20:10snorpJanH: ASAP
20:10snorpJanH: droeh has a patch that gets us pretty close
20:20* ahunt updated Android Studio hoping for less hangs and OOMs. It decided to freeze on first start...
17 May 2017
No messages
Last message: 94 days and 20 hours ago