1237823 - [Simulator] The Simulator can't resize for full screen on Mac.

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Simulator, defect)

Description

[Mantaroh Yoshinaga[:mantaroh]]

The simulator can't resize to full screen size on OSX.
The simulator repeated full screen and normal size.

Reproduce Step:
1. Launch the Firefox OS 2.6 Simulaotr on OSX.
2. Click the maximize button.

Actually Result:
 The simulator alternated between full screen size and normal size.

Expected Result:
 We can resize to full screen.

This bug reported by twitter(https://twitter.com/hikaru__m/status/685140804805853184).

Additional Information:
 - This phenomenon occurred on Max OS only.
 - This phenomenon occurred when click maximize button with option button.[1]

[1] http://osxdaily.com/2015/03/11/change-the-green-button-maximize-behavior-in-os-x-yosemite-with-bettertouchtool/

Comment 1

[Mantaroh Yoshinaga[:mantaroh]]

Hi Janx, 

I investigating this bug. Please let me ask the question about Simulator's window.

It is my understanding that Mulet and B2G-Desktop used shell.html as Chrome window. Am I right?

Comment 2

[Jan Keromnes [:janx]]

Hi Mantaroh,

Thank you for looking into this bug. I believe that both Mulet and B2G-Desktop use shell.html, but they probably use different chrome windows around that.

Alex should know more about this.

Comment 3

[Takayoshi]

I found this bug.

This environment is 
Mac OSX 10.10.5 Yosemite,
Firefox Developer Edition 45.0a2,
Simulator Firefox OS 2.2 or later(2.2, 2.6, 2.6_tv, (3.0)).

Comment 4

[Alexandre Poirot [:ochameau]]

(In reply to Mantaroh Yoshinaga[:mantaroh] from comment #1)
> Hi Janx, 
> 
> I investigating this bug. Please let me ask the question about Simulator's
> window.
> 
> It is my understanding that Mulet and B2G-Desktop used shell.html as Chrome
> window. Am I right?

Mulet uses browser.xul (the Firefox top level document), but b2g-desktop and simulators are using shell.html. Simulators are based on mulet, but overrides the top level document back to shell.html.
So that even if simulators are based on mulet, they act like b2g-desktop.

This bug most likely lives in this function:
  http://mxr.mozilla.org/mozilla-central/source/b2g/chrome/content/screen.js#193

Comment 5

[Mantaroh Yoshinaga[:mantaroh]]

Thanks janx, ochameau.

Sorry for late reply.
I can reproduce on Desktop using by the following code, in case of OSX.
----------------------
window.onresize = (msg) => {
  timeoutId = setTimeout(function() { 
    timeoutId = undefined; 
    window.resizeTo(window.outerWidth, window.outerHeight);
  },0);
};
----------------------
I don't know the reason yet.
I'll continue investigating this problem.

Comment 6

[Mantaroh Yoshinaga[:mantaroh]]

Attachment: bug1237823.mov

Perhaps, The reason of this phenomenon was reduced window width when resize to maximum window size.

Reproduce Step:
 1) Launch browser and go to "about:about".
 2) open the |Web Console|.
 3) execute the following scripts.
   3-1 ) window.onresize = (msg) => { console.log(window.outerWidth); };
   3-2 ) window.resizeTo(screen.availWidth, screen.availHeight);
   3-3 ) window.resizeTo(window.outerWidth, window.outerHeight);

Actually Result: 
  window width is dwindling.

Expected Result:
  window width is NOT dwindling.

Comment 7

[Mantaroh Yoshinaga[:mantaroh]]

Attachment: Not call resizeTo() when resize to same width as last time.

Comment 8

[Mantaroh Yoshinaga[:mantaroh]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=0bbf89bae7ea&exclusion_profile=false

Comment 9

[Mantaroh Yoshinaga[:mantaroh]]

I think that cocoa widget is identified maximized window size when step 3-2) of comment #6.
And cocoa widget zoomed when step 3-3 of comment #6 [1]. This condition is occurred by nsXULWindow::SetSize[2]. 

So We had better to change the gaia part in order to prevent this phenomenon.

[1] https://dxr.mozilla.org/mozilla-central/rev/584870f1cbc5d060a57e147ce249f736956e2b62/widget/cocoa/nsCocoaWindow.mm#1235
[2] https://dxr.mozilla.org/mozilla-central/rev/584870f1cbc5d060a57e147ce249f736956e2b62/xpfe/appshell/nsXULWindow.cpp#641

Comment 10

[Mantaroh Yoshinaga[:mantaroh]]

Attachment: Not call resizeTo() when resize to same width as last time.

rebased and added check-in comment.

Hi Alexandre,

Could you confirm this patch?
I would like to change screen.js in order to prevent endless loop.

Comment 11

[Alexandre Poirot [:ochameau]]

Comment on attachment 8714688 [details] [diff] [review]
Not call resizeTo() when resize to same width as last time.

Review of attachment 8714688 [details] [diff] [review]:
-----------------------------------------------------------------

I'm not peer of that code. But that looks good to me.
Please submit an updated patch and ask review to :fabrice.

::: b2g/chrome/content/screen.js
@@ +158,5 @@
> +
> +      if (lastResizedWidth == (width + chromewidth)) {
> +        return;
> +      }
> +      lastResizedWidth = (width + chromewidth);

Would that work with:
if (width+chromewidth == window.outerWidth) {
  return;
}
Or:
if (lastResizedWidth == width) {
  return;
}	
lastResizedWidth = width;

It would be better to use the first if that works.

Comment 12

[Mantaroh Yoshinaga[:mantaroh]]

Attachment: Not call resizeTo() when resize to same width as last time.

Thanks Alexandre.

(In reply to Alexandre Poirot [:ochameau] from comment #11)
> I'm not peer of that code. But that looks good to me.
> Please submit an updated patch and ask review to :fabrice.
OK.
I'll request to fabrice.

> ::: b2g/chrome/content/screen.js
> @@ +158,5 @@
> > +
> > +      if (lastResizedWidth == (width + chromewidth)) {
> > +        return;
> > +      }
> > +      lastResizedWidth = (width + chromewidth);
> 
> Would that work with:
> if (width+chromewidth == window.outerWidth) {
>   return;
> }
> Or:
> if (lastResizedWidth == width) {
>   return;
> }	
> lastResizedWidth = width;
> 
> It would be better to use the first if that works.
I guess that we had better to choice second.
The outerWidth same value of |width+chromewidth| already this statement.


Hi fabrice.
I work on Mulet resizing problem on Mac OSX.
Could you confirm this patch?

Comment 13

[Mantaroh Yoshinaga[:mantaroh]]

Thanks fabrics.

Comment 14

[Pulsebot]

https://hg.mozilla.org/integration/b2g-inbound/rev/90ead10ecd0e

Comment 15

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/90ead10ecd0e