824386 - Fullscreen elements are larger than the dimensions of the display on high DPI displays

Status: RESOLVED FIXED

(Core :: Widget, defect)

Description

[Josh Tumath]

User Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:20.0) Gecko/20121223 Firefox/20.0
Build ID: 20121223030832

Steps to reproduce:

1. In the Control Panel, set the DPI of the desktop to 150%.
2. Enable the HTML5 video player on YouTube
3. Open a YouTube video that uses the HTML5 video player and open it in fullscreen


Actual results:

The dimensions of the fullscreen element are about 30px than the actual size of the display, so the controls at the bottom of the YouTube video player are slightly cut off.


Expected results:

The fullscreen element should be the same dimensions as the resolution of the display.

Comment 1

[Greg Edwards]

Mozilla/5.0 (Windows NT 6.1; WOW64; rv:20.0) Gecko/20130114 Firefox/20.0

I can replicate on Windows 7 with 200% DPI using the built-in fullscreen option in the main menu. It affects all pages, and the most notable effect is the scrollbar on the right edge is cut off. My devPixelsPerPx is manually set to -1.0.

This should block Bug 820679.

Comment 2

[Jonathan Kew [:jfkthame]]

I also notice that when the window is restored from full-screen to "normal", it does not return to the correct original size. I suspect this is an aspect of the same underlying problem with handling the window dimensions when scaling is in effect.

Comment 3

[Gary [:streetwolf52]]

I see this also using 125% (120DPI). F11 full zoom extends to my second monitor about 25%.  Also various menus opened in Fx and initially moved to my second monitor will always open about 1/4 of the way from the left margin. View Page Info does this as well as other things.

Comment 4

[Jim Mathies [:jimm]]

Attachment: screenshot

Comment 5

[Jim Mathies [:jimm]]

This should probably track since it's not just visual polish.

Comment 6

[Gary [:streetwolf52]]

Just wanted to add that the problem I have with menus and windows aligning themselves on my second monitor 25% from the left side happens no matter what window size Fx is in. It's not limited to full screen mode (F11).

Comment 7

[Jim Mathies [:jimm]]

(In reply to Gary [:streetwolf] from comment #6)
> Just wanted to add that the problem I have with menus and windows aligning
> themselves on my second monitor 25% from the left side happens no matter
> what window size Fx is in. It's not limited to full screen mode (F11).

Please file a new bug blocking bug win-hidpi with a screen shot. This bug is specific to dom fullscreen video controls. If we can get individual bugs filed like this one on each individual problem we'll have much better luck finding owners who can fix the fallout.

Comment 8

[Gary [:streetwolf52]]

Opened https://bugzilla.mozilla.org/show_bug.cgi?id=851952

Comment 9

[Gary [:streetwolf52]]

Jim... is there one place to find all the reported bugs due to win-hidpi?  I have issues with the way the chrome looks and blurry graphics in chrome and content.  I recall reading that these are already being worked on.

Comment 10

[Chris Pearce [:cpearce (Not reading bugmail)]]

Hmm... I cannot reproduce this. Maybe it's a layout issue rather than a fullscreen issue though.

Comment 11

[Jim Mathies [:jimm]]

(In reply to Gary [:streetwolf] from comment #9)
> Jim... is there one place to find all the reported bugs due to win-hidpi?  I
> have issues with the way the chrome looks and blurry graphics in chrome and
> content.  I recall reading that these are already being worked on.

The parent of this bug - bug 824386.

Comment 12

[Jim Mathies [:jimm]]

(In reply to Chris Pearce (:cpearce) from comment #10)
> Hmm... I cannot reproduce this. Maybe it's a layout issue rather than a
> fullscreen issue though.

My str:

1) on the surface pro, open nightly on the desktop
2) visit http://www.w3.org/2010/05/video/mediaevents.html
3) click the fullscreen button on the bottom right

Comment 13

[Jim Mathies [:jimm]]

(In reply to Jim Mathies [:jimm] from comment #11)
> (In reply to Gary [:streetwolf] from comment #9)
> > Jim... is there one place to find all the reported bugs due to win-hidpi?  I
> > have issues with the way the chrome looks and blurry graphics in chrome and
> > content.  I recall reading that these are already being worked on.
> 
> The parent of this bug - bug 824386.

sorry - copy paste error - bug 820679

Comment 14

[Chris Pearce [:cpearce (Not reading bugmail)]]

(In reply to Jim Mathies [:jimm] from comment #12)
> (In reply to Chris Pearce (:cpearce) from comment #10)
> > Hmm... I cannot reproduce this. Maybe it's a layout issue rather than a
> > fullscreen issue though.
> 
> My str:
> 
> 1) on the surface pro, open nightly on the desktop
> 2) visit http://www.w3.org/2010/05/video/mediaevents.html
> 3) click the fullscreen button on the bottom right

Ah, I was using Nightly from Feb 28, I updated to today's (17 Mar) and I can now reproduce the bug on my Surface Pro.

So this must be a recent regression, sometime between 2013-02-28 and 2013-03-17.

Comment 15

[Jonathan Kew [:jfkthame]]

It will have been triggered by bug 844604 (landed 2013-03-14), where we enabled scaling based on the Windows resolution setting.

(My guess is that we're working with the physical pixel dimensions of the screen, but sizing the window using logical coordinates, or something like that - haven't tried to dig in properly yet.)

Comment 16

[Gary [:streetwolf52]]

(In reply to Jim Mathies [:jimm] from comment #12)
> (In reply to Chris Pearce (:cpearce) from comment #10)
> > Hmm... I cannot reproduce this. Maybe it's a layout issue rather than a
> > fullscreen issue though.
> 
> My str:
> 
> 1) on the surface pro, open nightly on the desktop
> 2) visit http://www.w3.org/2010/05/video/mediaevents.html
> 3) click the fullscreen button on the bottom right

When I do this I get what I describe in my bug report regarding full screen mode.
https://bugzilla.mozilla.org/show_bug.cgi?id=851952

Comment 17

[Alex Keybl [:akeybl]]

From our read, bug 820679 hasn't been resolved yet. No need to track until there's actually a functional regression in 22.

Comment 18

[Jonathan Kew [:jfkthame]]

Actually, at this point there -is- a functional regression in 22 as a result of bug 844604, as per comments 14 and 15.

I'm expecting to have a patch ready for this shortly, but if for some reason that proves problematic, I think it would be sufficient reason to hold 844604 back for a cycle.

Comment 19

[Jonathan Kew [:jfkthame]]

Attachment: patch, account for logical dpi in nsScreenWin

This should also fix the issues with incorrectly centered windows (bug 844857) and failing to constrain restored windows to the screen size (bug 852371).

Comment 20

[Jim Mathies [:jimm]]

Comment on attachment 726569 [details] [diff] [review]
patch, account for logical dpi in nsScreenWin

Review of attachment 726569 [details] [diff] [review]:
-----------------------------------------------------------------

::: widget/windows/nsScreenWin.cpp
@@ +93,5 @@
> +                               int32_t *outWidth, int32_t *outHeight)
> +{
> +  int32_t left, top, width, height;
> +  nsresult rv = GetRect(&left, &top, &width, &height);
> +  if (NS_SUCCEEDED(rv)) {

nit - I'd prefer 

if (NS_FAILED(rv))
  return rv;
...
return NS_OK;

Comment 21

[Jonathan Kew [:jfkthame]]

OK, pushed with nits adjusted:
https://hg.mozilla.org/integration/mozilla-inbound/rev/d37c97ba174f

Comment 22

[Greg Edwards]

What are the chances that your patch will also fix Bug 818927? (seems to involve absolute mouse positions -> css pixels)

Comment 23

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/d37c97ba174f