The task switcher is in a pretty bad shape. I see a black "System" app show up in the task switcher. It sometimes is black. Once I also saw the lockscreen in that screenshot at some point. To make matters worse, it was about half height (square) and the lockscreen was rendered wrong inside of it. Maybe an orientation issue? I have seen the small/square misrendered snapshots with other apps too, especially the settings app. I have no clear STR, but I think the system app has something to do with the permissions prompt or maybe E.me. I was using the maps app and it keeps asking me for GPS permissions which I allow. I also opened an E.me app previously. While I have no clear STR, I have seen this symptoms very frequently so it shouldn't be too hard to trigger it. This is utter brokenness and a dogfooding blocker I believe.

System app should never appear on that. If that happens, it's a bug. We need to find out what is causing System being launched by window manager. Screenshot of Settings app can be small, because we launched it in background before style sheets is loaded as we received first system message from SIM tool kit. Permission prompt has nothing to do with this. E.me (wrapper) might, i'll pass that question to vivien.

I think I know who launch the System app! Taking.

(In reply to Andreas Gal :gal from comment #0) > The task switcher is in a pretty bad shape. > > I see a black "System" app show up in the task switcher. It sometimes is > black. Once I also saw the lockscreen in that screenshot at some point. To > make matters worse, it was about half height (square) and the lockscreen was > rendered wrong inside of it. Maybe an orientation issue? > This is a side effect of a patch to fix some crashes with system messages. > I have seen the small/square misrendered snapshots with other apps too, > especially the settings app. > This happens when an application is opened in the background with a system message. It is not size correctly at this time and we take a screenshot. I believe we should not.

Why are we auto launching settings?

Half-size apps is being dealt in bug 804396. Patch to come, few liner fix.

(In reply to Vivien Nicolas (:vingtetun) from comment #3) > > This is a side effect of a patch to fix some crashes with system messages. > I think it's because of bug 796670 -- when we received the alarm system message, window manager mistakenly thinks that it should launch the system app. This is a one liner fix, Patch to come.

https://github.com/mozilla-b2g/gaia/pull/5973 Testing STR: 1. Connect Otoro to wifi network 2. Disconnect USB and turn off the screen, wait for > 60 sec 3. Wifi will be quietly turned off at 60th second by a alarm API system message 4. Turn on the screen. wifi will briefly seem disconnected, but soon reconnects. 5. Go to task switcher and verify the same system message doesn't launch the System app in a frame. I will expect with the patch you will not see the System app; without you will see it.

Are any of the scenarios here covered by our smoketests?

Tim, why are we launching the settings app here?

Tim, please answer comment #10. We probably want a follow-up bug for that.

(In reply to Andreas Gal :gal from comment #10) > Tim, why are we launching the settings app here? When the phone is connected to cellar network, RIL will dispatch a "icc-stkcommand" system message for Settings app to handle the SIM Tool Kit menu. System message will dispatch a generic 'open-app' ChromeEvent to System app first, so we can launch the app that handles the message.

Sorry, still not getting it. Why is the settings app started here? What is that menu for? What does the settings app do in response to that command?

(In reply to Andreas Gal :gal from comment #13) I also CC philikon and frsela (who implemented STK gaia) > Why is the settings app started here? > What is that menu for? What does the settings app do in response to that command? When SIM_READY, SIM card will send a STK command(icc-stkcommand) In current UX design, STK menu is in Settings App, so Settings App will be launched to handle this STK command. Usually this command consists of "STK Menu"(title, items), so Settings will need to handle this command to show the STK Menu accordingly.

(In reply to Yoshi Huang[:yoshi][:allstars.chh] from comment #14) > (In reply to Andreas Gal :gal from comment #13) > I also CC philikon and frsela (who implemented STK gaia) > > > Why is the settings app started here? > > What is that menu for? What does the settings app do in response to that command? > When SIM_READY, SIM card will send a STK command(icc-stkcommand) > In current UX design, STK menu is in Settings App, > so Settings App will be launched to handle this STK command. > > Usually this command consists of "STK Menu"(title, items), > so Settings will need to handle this command to show the STK Menu > accordingly. Is this related with https://bugzilla.mozilla.org/show_bug.cgi?id=800274?

Can someone clarify if this gaia merge was landed into the Unagi daily build that was generated at 24-Oct-2012 10:35 ? https://github.com/mozilla-b2g/gaia/pull/5980 seems to indicate that it was in the merge batch, and the daily build was created much after this happened. However, I'm testing on that 10-24 daily build, and i still see the rocket App, and clicking it will trigger bug 804883, meanwhile, long tapping Home will bring up the System menu in the window manager. I fear that not seeing this fix will break another update to stable.

(In reply to Tony Chung [:tchung] from comment #16) > Can someone clarify if this gaia merge was landed into the Unagi daily build > that was generated at 24-Oct-2012 10:35 ? > > https://github.com/mozilla-b2g/gaia/pull/5980 seems to indicate that it was > in the merge batch, and the daily build was created much after this happened. > > However, I'm testing on that 10-24 daily build, and i still see the rocket > App, and clicking it will trigger bug 804883, meanwhile, long tapping Home > will bring up the System menu in the window manager. > > I fear that not seeing this fix will break another update to stable. And i'll clarify that i'm still seeing the STR in comment 7 with the daily build.

(In reply to Yoshi Huang[:yoshi][:allstars.chh] from comment #14) > (In reply to Andreas Gal :gal from comment #13) > I also CC philikon and frsela (who implemented STK gaia) > > > Why is the settings app started here? > > What is that menu for? What does the settings app do in response to that command? > When SIM_READY, SIM card will send a STK command(icc-stkcommand) > In current UX design, STK menu is in Settings App, > so Settings App will be launched to handle this STK command. > > Usually this command consists of "STK Menu"(title, items), > so Settings will need to handle this command to show the STK Menu > accordingly. First of all, the settings app doesn't have to handle this. We can let the system app handle this and for the rare case that anyone cares to invoke the STK menu, the settings app can pick up the last stored event from there. In any case, even if the settings app handles this, there is no way on earth this activity should have the same entry point and load dozens and dozens of resources as the main settings entry point. Can we at least make a separate tiny entry point file? Still though, option 1 is preferred. Not starting anything at all and letting the system app hold on to the event until its needed. Startup is getting slow and this is making it substantially worse.

Please figure out comment 16/17.

tchung, what's your gaia commit info?

Let's work on stk / wifi in a followup bug please.

I'm still unable to reproduce with a build from gaia d44b7a5b1b0d15e9bfc12e0476059afe3c247e08 / inbound d408a2592685.

I can reproduce this 100% on the parent commit of https://github.com/mozilla-b2g/gaia/pull/5980 , and reproduce 0% after that commit. tchung, let's check if your build has an issue. If not, we may be seeing a separate bug :/. This specific bug as testable is fixed.

(In reply to Chris Jones [:cjones] [:warhammer] from comment #20) > tchung, what's your gaia commit info? gecko: f58edfde05cb708f8a2c440d338f2e429aaf372b gaia: fcfa1857bed6596e992263206451c6814e4b2f6d

just adding a screenshot of the card view

scratch that last comment, it's a non-dogfood blocker. this fix has been verified on the 10-24 daily unagi build.