Okay, that's cool. We now know it isn't a manufacturer quirk it is an environment quirk.
It's not actually an inheritance problem. The order of methods being called ends up being: onKeyDown onKeyUp onBackPressed However, if your Activity overrides onKeyUp and it does not caller super.onKeyUp() when it is handling the back key the method onBackPressed() will never be called. It is the super.onKeyUp() that calls onBackPressed(). That was what was going on when I was using your back key changes. Simon Mac Donald http://hi.im/simonmacdonald On Thu, Apr 19, 2012 at 3:32 PM, Joe Bowser <[email protected]> wrote: > OK, here's what happened. > > When I currently work on Cordova, I run the project as an activity, and not > as a jar. This is done for convenience because I've had jars lose sync > with the code that I'm debugging in Eclipse in the past. However, what > this means is that I don't have a working back button. > > Activity has a default behaviour for the back button. This default > behaviour is defined by onBackPressed and onKeyDown when you extend an > activity, which is of course that the application would exit. Therefore, > one would assume that since any class that uses the jar extends DroidGap > and that since DroidGap is extending Activity that the class would extend > Activity by virtue of it being a DroidGap class. > > In reality this is wrong, and somehow through some voodoo the onKeyUp event > work even though it shouldn't because the App should have ran the > Activity's onKeyDown and onBackPressed. > > In short, I have to use the bin/create scripts because Java inheritance > doesn't seem to work the way I thought it should. > > Joe > > On Thu, Apr 19, 2012 at 12:22 PM, Simon MacDonald < > [email protected] > > wrote: > > > That's ominous. > > > > Simon Mac Donald > > http://hi.im/simonmacdonald > > > > > > On Thu, Apr 19, 2012 at 3:20 PM, Joe Bowser <[email protected]> wrote: > > > > > OK, let's revert the changes. It looks like I'm going to have to > change > > > how I work on this project. > > > > > > On Thu, Apr 19, 2012 at 12:15 PM, Filip Maj <[email protected]> wrote: > > > > > > > Actually the latest master breaks back button on my galaxy nexus too. > > > > > > > > I'll run through the tests and help out with this. > > > > > > > > On 4/19/12 12:07 PM, "Joe Bowser" <[email protected]> wrote: > > > > > > > > >There's a problem with the log statement that you provided, namely > the > > > > >fact > > > > >that this.backHistory() actually goes back in the history before it > > > > >returns > > > > >true or false. This was causing the app to constantly exit when we > > were > > > > >running mobile-spec, which was similar to the original bug. I > > removed > > > it > > > > >because I wanted the log statements to just log, and here's what I > got > > > > >from > > > > >the Motorola Device: > > > > > > > > > >04-20 02:56:14.187 19771 19771 D BackPressed: in onBackPressed > > > > >04-20 02:56:14.187 19771 19771 D BackPressed: bound = false > > > > >04-20 02:56:14.187 19771 19771 D BackPressed: exiting > > > > >04-20 02:56:14.234 1032 1032 D ActivitityRenderTarget: onResume > > > > >04-20 02:56:14.257 19771 19771 D BackButton: I got an up from > > > KEYCODE_BACK > > > > >04-20 02:56:14.265 19771 19771 D BackButton: I'm doing nothing > > > > >04-20 02:56:14.265 19771 19771 D BackPressed: returning false > > > > > > > > > >So, here we're exiting the app on the keyDown, and the keyUp is also > > > > >happening which is exactly what I would expect since you have to do > a > > > > >onKeyDown before an onKeyUp. Here's the same thing on the Samsung > > > Galaxy > > > > >S2: > > > > > > > > > >D/BackPressed(17630): in onBackPressed > > > > >D/BackPressed(17630): bound = false > > > > >D/BackPressed(17630): exiting > > > > >I/InputDispatcher( 2704): Delivering key to current input target: > > > action: > > > > >1, channel '409de9f8 > org.apache.cordova/org.apache.cordova.StandAlone > > > > >(server)' > > > > > > > > > >...A bunch of crap about widgets not being drawn right on the home > > > screen > > > > >which is resuming... > > > > > > > > > >D/BackButton(17630): I got an up from KEYCODE_BACK > > > > >D/BackButton(17630): I'm doing nothing > > > > >D/BackPressed(17630): returning false > > > > > > > > > >I should remove the crap Samsung puts on their default build, but > this > > > > >does > > > > >illustrate my point more clearly. Once we remove the extra > > backHistory > > > > >call, things seem to work properly on this end and the app does > exit. > > > It > > > > >seems incredibly strange that the Note would fail. > > > > > > > > > >On Thu, Apr 19, 2012 at 11:48 AM, Simon MacDonald > > > > ><[email protected] > > > > >> wrote: > > > > > > > > > >> Oh, and here is the dirt simple .html I was using. I was checking > a > > > > >> possible camera bug and found I couldn't exit the app. > > > > >> > > > > >> http://dl.dropbox.com/u/887989/camera.html > > > > >> > > > > >> Simon Mac Donald > > > > >> http://hi.im/simonmacdonald > > > > >> > > > > >> > > > > >> On Thu, Apr 19, 2012 at 2:44 PM, Simon MacDonald > > > > >> <[email protected]>wrote: > > > > >> > > > > >> > You can get my version of DroidGap with the extra logs from: > > > > >> > > > > > >> > http://dl.dropbox.com/u/887989/DroidGap.java > > > > >> > > > > > >> > Did you clone the 1.6.1 tag or the bleeding edge? > > > > >> > > > > > >> > Simon Mac Donald > > > > >> > http://hi.im/simonmacdonald > > > > >> > > > > > >> > > > > > >> > On Thu, Apr 19, 2012 at 2:38 PM, Joe Bowser <[email protected]> > > > > wrote: > > > > >> > > > > > >> >> I cloned a fresh repo, and what's in DroidGap.java works on > this > > > end, > > > > >> and > > > > >> >> matches what I have in my working tree. We don't have your > > > attached > > > > >> >> DroidGap.java, can you send it again? > > > > >> >> > > > > >> >> On Thu, Apr 19, 2012 at 11:28 AM, Simon MacDonald < > > > > >> >> [email protected] > > > > >> >> > wrote: > > > > >> >> > > > > >> >> > When I pull the latest from incubator-cordova-android and > build > > > it > > > > >> into > > > > >> >> my > > > > >> >> > app the back key does not work. I've verified this on my > > samsung > > > > >>note > > > > >> >> and > > > > >> >> > on the emulator. > > > > >> >> > > > > > >> >> > Please take a look at the attached DroidGap.java file which > > I've > > > > >> added a > > > > >> >> > few logs. When I click on the back button on the note or on > the > > > > >> >> emulator I > > > > >> >> > see these logs: > > > > >> >> > > > > > >> >> > 04-19 14:16:49.373: D/BackButton(390): I got an up from > > > > >>KEYCODE_BACK > > > > >> >> > 04-19 14:16:49.373: D/BackButton(390): I'm doing nothing > > > > >> >> > 04-19 14:16:49.373: D/BackPressed(390): returning false > > > > >> >> > > > > > >> >> > I'm not seeing the onBackPressed method being called at all. > > Can > > > > >> Joe/Fil > > > > >> >> > try my DroidGap.java to see if they get a different set of > > logs? > > > > >> >> > > > > > >> >> > Note that the .html I'm loading has one and only one file and > > > does > > > > >>not > > > > >> >> > call 'document.addEventListener("backbutton", onBackKey, > > true)". > > > > >>It is > > > > >> >> just > > > > >> >> > a plain vanilla .html file. > > > > >> >> > > > > > >> >> > Simon Mac Donald > > > > >> >> > http://hi.im/simonmacdonald > > > > >> >> > > > > > >> >> > > > > > >> >> > On Thu, Apr 19, 2012 at 2:17 PM, Filip Maj <[email protected]> > > > wrote: > > > > >> >> > > > > > >> >> >> We definitely cannot revert that commit Joe put in. Back > > button > > > > >> didn't > > > > >> >> >> work at all (just exited) before that commit. > > > > >> >> >> > > > > >> >> >> Sucks that the note is broken but it works as expected on > > other > > > > >> devices > > > > >> >> >> (as Joe points out). > > > > >> >> >> > > > > >> >> >> Simon can you try to debug the issue on the Note and see how > > the > > > > >> >> execution > > > > >> >> >> flow differs from the other Android devices? > > > > >> >> >> > > > > >> >> >> On 4/19/12 10:58 AM, "Joe Bowser" <[email protected]> > wrote: > > > > >> >> >> > > > > >> >> >> >OK, I just went through Mobile Spec's manual tests on the > > > > >>following > > > > >> >> >> >devices: > > > > >> >> >> > > > > > >> >> >> >Galaxy Nexus > > > > >> >> >> >Nexus S > > > > >> >> >> >Motorola RAZR > > > > >> >> >> >Samsung Galaxy S2 > > > > >> >> >> >LG Optimus One > > > > >> >> >> > > > > > >> >> >> >Currently, if you don't override the back button, the > > > backPressed > > > > >> >> event > > > > >> >> >> >fires on the keyDown. If we override the event, the > > backbutton > > > > >> >> doesn't > > > > >> >> >> >work on the keydown, but the event fires on the keyup > > instead. > > > > >> This I > > > > >> >> >> >think is the proper behaviour since we're intercepting the > > back > > > > >> button > > > > >> >> >> and > > > > >> >> >> >making it the users' responsibility to handle this event. > > > > >> >> >> > > > > > >> >> >> >The old method had the default behaviour for the back > button > > > > >>(where > > > > >> it > > > > >> >> >> >quits on keydown) occur, as well as the keyUp behaviour > > happen. > > > > >>Of > > > > >> >> >> >course, > > > > >> >> >> >since the app is exited and is in the background whatever > > > happens > > > > >> with > > > > >> >> >> the > > > > >> >> >> >back button at that point is moot. The fact is that I > expect > > > the > > > > >> back > > > > >> >> >> >button to work when I press it, and I don't consciously > think > > > > >>about > > > > >> >> key > > > > >> >> >> up > > > > >> >> >> >or key down when I do so. I also think that Google put it > on > > > > >> keydown > > > > >> >> for > > > > >> >> >> >a > > > > >> >> >> >reason, especially since it's not really a key anymore. > > > > >> >> >> > > > > > >> >> >> >On Thu, Apr 19, 2012 at 10:48 AM, Joe Bowser < > > > [email protected]> > > > > >> >> wrote: > > > > >> >> >> > > > > > >> >> >> >> You're saying that the behaviour is different on the > Note? > > > I'll > > > > >> run > > > > >> >> >> >> through our device bank here, but the fact is that the > Back > > > > >>Button > > > > >> >> >> would > > > > >> >> >> >> ALWAYS exit, not progress through the history but just > > > > >>straight up > > > > >> >> exit > > > > >> >> >> >> regardless of where in the app you were. What I added > was > > > > >> >> backPressed > > > > >> >> >> >> which would only exit if you were at the top of the > > history. > > > > >> Also, > > > > >> >> >> >> backPressed should be supported for Android versions > higher > > > > >>than > > > > >> >> >> Android > > > > >> >> >> >> 2.0, and since we're only supporting 2.1+, we should have > > > been > > > > >> fine. > > > > >> >> >> >> > > > > >> >> >> >> I'll investigate this, but honestly if it's not > consistent > > > > >>across > > > > >> >> >> >>devices, > > > > >> >> >> >> what do we do then? > > > > >> >> >> >> > > > > >> >> >> >> On Thu, Apr 19, 2012 at 10:26 AM, Simon MacDonald < > > > > >> >> >> >> [email protected]> wrote: > > > > >> >> >> >> > > > > >> >> >> >>> The reason we switched to keyUp is so you could slide > your > > > > >>finger > > > > >> >> off > > > > >> >> >> >>>the > > > > >> >> >> >>> up button in case you changed your mind about clicking > on > > > the > > > > >> back > > > > >> >> >> >>>button. > > > > >> >> >> >>> > > > > >> >> >> >>> I noticed that you checked in: > > > > >> >> >> >>> > > > > >> >> >> >>> > > > > >> >> >> >>> > > > > >> >> >> >>> > > > > >> >> >> > > > > >> >> > > > > >> > > > > https://git-wip-us.apache.org/repos/asf?p=incubator-cordova-android.git > > > > >> ; > > > > >> >> >> >>>a=commit;h=180696baec28d6e6b67e8bb23a5302f21bce6b79 > > > > >> >> >> >>> > > > > >> >> >> >>> On my Samsung Galaxy Note running Android 2.3.6 the back > > > > >>button > > > > >> >> does > > > > >> >> >> >>>not > > > > >> >> >> >>> work. That is when I do not regisiter for a "backbutton" > > > > >>listener > > > > >> >> and > > > > >> >> >> I > > > > >> >> >> >>> expect that when I click the back key that my app would > > > exit. > > > > >> With > > > > >> >> our > > > > >> >> >> >>>fix > > > > >> >> >> >>> the back button doesn't exit the app. I have to click > the > > > home > > > > >> >> button > > > > >> >> >> >>>to > > > > >> >> >> >>> exit. > > > > >> >> >> >>> > > > > >> >> >> >>> Reverting your change returns the back button behaviour > to > > > > >> normal. > > > > >> >> So > > > > >> >> >> >>> something is still janky. I haven't been able to test it > > on > > > my > > > > >> >> phone > > > > >> >> >> >>>as I > > > > >> >> >> >>> can't convince it to be recognized on my laptop. > > > > >> >> >> >>> > > > > >> >> >> >>> Simon Mac Donald > > > > >> >> >> >>> http://hi.im/simonmacdonald > > > > >> >> >> >>> > > > > >> >> >> >>> > > > > >> >> >> >>> On Tue, Apr 17, 2012 at 8:46 PM, Joe Bowser > > > > >><[email protected]> > > > > >> >> >> wrote: > > > > >> >> >> >>> > > > > >> >> >> >>> > Hey > > > > >> >> >> >>> > > > > > >> >> >> >>> > Remember this issue: > > > > >> >> >> >>> > https://issues.apache.org/jira/browse/CB-367 > > > > >> >> >> >>> > > > > > >> >> >> >>> > I'm surprised that I'm catching it now, but the Back > > > Button > > > > >>is > > > > >> >> >> really > > > > >> >> >> >>> janky > > > > >> >> >> >>> > on my Nexus S running 2.3.6, and I'm working on a fix > > now. > > > > >> What > > > > >> >> was > > > > >> >> >> >>>the > > > > >> >> >> >>> > reasoning behind the Back Button being on the keyUp > > > instead > > > > >>of > > > > >> >> the > > > > >> >> >> >>> keyDown? > > > > >> >> >> >>> > I don't remember. > > > > >> >> >> >>> > > > > > >> >> >> >>> > Joe > > > > >> >> >> >>> > > > > > >> >> >> >>> > > > > >> >> >> >> > > > > >> >> >> >> > > > > >> >> >> > > > > >> >> >> > > > > >> >> > > > > > >> >> > > > > >> > > > > > >> > > > > > >> > > > > > > > > > > > > > >
