I cannot repro this on Chrome M51 on Mac if I go fullscreen. Though if I do switch to another display, and click on a Chrome window, I do in fact get the bookmarks bar showing up.  though this might be a completely different issue.

assigning to yining to repro.

i've seen this recently on M52 canary, win7 i think.  i'll see if it can repro it and provide details if so.

yes, i can repro it.  win7 home premium, m52 canary.  this was a 100% repro for me:

1. start browser
2. "html5 video" in omnibox
3. select w3cschools
4. click full screen
5. observe task bar.

exiting and re-entering full screen does not fix it.  to fix it, the only way i found was to have the browser window lose focus.  then full screen starts working.  i switched focus by clicking on the desktop while the w3c page was loaded, not in full screen.  didn't try other variations on this (e.g., alt-tab).

i'll attach a screen shot tomorrow morning.

I have the same issue on linux since I've upgraded from chromium 49 to 50.

Can reproduce issue on Windows 10 Beta Channel 51.0.2704.36

See the attachment 

Deleted: Untitled.jpg 137 KB

	Untitled.jpg 137 KB View Download

+mgiuca who has fixed some fullscreen issues recently and might know who to route this too.

I can't repro this on Windows, Chrome Stable 50.0.2661.87 (which is very close to the version reported). Nor on Windows, Chrome Canary 52.0.2714.0. On both YouTube and w3schools, the fullscreen button on the video correctly enters fullscreen and no OS or browser UI is visible.

To people having this issue: when you click the fullscreen button, do you see the black message "Press [Esc] to exit fullscreen"? i.e., is it working at all or does it just do nothing?

#11:
> 1. start browser
> 2. "html5 video" in omnibox
> 3. select w3cschools

It is preferable in a bug report if you just provide the URL of the page in question rather than instructions for how to find it in a Google search. I assume you refer to this page:
http://www.w3schools.com/html/html5_video.asp

Also renaming the issue as it does not seem to be related to the taskbar, but rather fullscreen is not working at all.

I found a post on Reddit about this with a number of responses indicating they also had the bug.
https://www.reddit.com/r/chrome/comments/4hvmfu/youtube_not_going_fullscreen_only_extending_to/

Also, can anyone experiencing this please try https://permission.site/ --- click Fullscreen and report whether it hides the Chrome browser UI and taskbar? I'm trying to figure out whether this is specific to videos or fullscreen in general.

Re #15, at least for me, there is no fullscreen warning message displayed.

miguaca: It depends on the tab. I currently have the site open in two tabs: In one of them it works as expected and I get the "... is now full screen |allow| |exit full screen|" message. In the other one nothing happens. The one in which is doesn't work is an older tab (see #3).

I've attached a video that shows the behavior (or lack thereof) of the fullscreen button both tabs.

Deleted: fs.gif 209 KB

	fs.gif 209 KB View Download

#18 That means the API itself is probably broken.

#19/#20: That video is very hard to watch, it jumps around everywhere and everything flashes so fast I can't tell what's going on, or even what operating system you have. Would you (or anyone else) be able to make a video of the entire screen in an actual video file (not a .gif)?

OK, I have personally tried:
Windows 7 32-bit (single and dual monitor)
Windows 10 64-bit (single monitor)

Still cannot reproduce this at all, for any YouTube videos (including ones that the Reddit posters are telling me to try), or https://permission.site.

It seems like this suddenly started happening a few days ago affecting all distros of Chrome, which makes me suspect it is a bug in something else (Windows, video card, YouTube). The fact that #18 says there is no fullscreen notice makes me suspicious that it is a system issue. And the fact that it happens on permission.site indicates it's not a YouTube issue.

As I mentioned above, I'm using linux. I'm using the open source intel
graphics drivers.

#23: I've tried Linux as well; no repro. I had a look at your GIF frame-by-frame. One thing that's odd is that it still has the old white "<site> has gone fullscreen" bubble, which we replaced in Chrome 49. Have you manually disabled the new bubble in chrome://flags? (I'm not sure how that would affect it not working but it's an interesting thing nonetheless.)

Do those other buttons on permission.site work? How about pointer lock?

The only change I remember making in //flags is disabling smooth scrolling. I've made another video that shows the full screen and also shows that pointer lock works.

Deleted: test.mkv 518 KB

test.mkv 518 KB Download

Thanks for making that video. I'm still puzzled about a few things:
1. In the video when you click Fullscreen on the left, it jumps to the top of the page and you appear to click the HTTPS flipper. Do you know what caused that?
2. Why do you not have the new black fullscreen bubble (i.e. https://bugs.chromium.org/p/chromium/issues/attachment?aid=221232&inline=1). Can you check chrome://version and report what version it says? Can you check chrome://flags/#simplified-fullscreen-ui and check what the simplified fullscreen flag is set to?

1. When I click the button it jumps to the top and by chance my mouse
cursor is hovering over the HTTPS button which causes the animation
you can see. I don't know whether jumping to the top in fullscreen is
part of the sites logic or whether chromium thinks that the viewport
size has changed and jumps to the top itself.

2. I do remember seeing the black bubble previously. I don't know
whether I've seen it only in 49 or also in 50. I've checked the flags
page and it appears that I've disabled the simplified fullscreen UI. I
don't remember doing this but I remember thinking that the new UX was
worse than the old one. So I probably changed it at the same time I
disabled smooth scrolling.

Chromium: 50.0.2661.75 (Developer Build) (64-bit)
Revision: f6c0a1a39f3cc347ffaf4a575be3ff53f270c7e6

This has happening to me since I upgraded to v50 from v49. Please fix this, this is so annoying.

Netscape

Web browser version: 5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/50.0.2661.94 Safari/537.36

Operating platform: Win32

User agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/50.0.2661.94 Safari/537.36 

I recorded video. See the video for the issue

https://drive.google.com/file/d/0B0B5Ac90yT1nR0NUNmtnV21SSjA/view

I can reproduce on Windows 10 and Chrome 50.0.2661.94 m

#27: Can you try setting simplified fullscreen UI flag back to Default and see if it's still an issue. We don't provide support for Chrome with any flags changed (and it would be helpful to know of any other flags you've changed).

#29: Thanks for recording a video. It does seem to be a regression from M49 to M50.

+ananta, since you are involved in Issue 604359, perhaps you know something about this. Could r377084 also be responsible for this regression? (It happened in the same time window.) NB I can't repro this one so I can't do a bisect.

+watk: I've seen you landed a number of CLs in the 49--50 time frame that mention fullscreen video. No clear reason why these would be responsible for the regression (particularly because this seems to affect non-video content), but perhaps you can think of something.

Can reproduce on Win10 with Chrome 50.0.2661.94 m

I've got an Android version of this bug (or one very similar) assigned to me (http://crbug.com/607128). Should I dupe it into this one? My changes should all be Android specific so I'm pretty sure it wasn't one of my changes.

#35: Let's not dupe just yet, given that the Android one apparently regressed in M52.

I have a job for anyone who's game to try out some custom builds of Chromium (for people who are able to reliably reproduce this issue). I want to test whether r377084 caused this issue.

If you want to help out, do these steps:
1. Go here:
https://commondatastorage.googleapis.com/chromium-browser-snapshots/index.html?prefix=Win/377079/
2. Download chrome-win32.zip.
3. Unzip and run chrome.exe.
4. Try to trigger the bug (e.g., go YouTube, click fullscreen, see if it works a couple of times).
5. If fullscreen *does* work, then I need you to test a newer version to make sure you can still reproduce the bug on a later build of Chromium. Go here:
https://commondatastorage.googleapis.com/chromium-browser-snapshots/index.html?prefix=Win/377106/
6. Follow the same steps as above.
7. If fullscreen *works* in both of those versions, go here:
https://commondatastorage.googleapis.com/chromium-browser-snapshots/index.html?prefix=Win/378086/

Report here whether fullscreen works or does not work in all of the versions you tried.

Also, for anyone reporting technical details here, it would be good to know:
- Operating system version (e.g., Windows 10).
- Chrome version (e.g., 50.0.2661.94, see chrome://version).
- 32 or 64 bit Chrome (see chrome://version).
- Number of monitors (I wonder if dual monitor is related).
- Graphics card manufacturer and model (e.g., Nvidia GeForce GTX 970).
- If you have disabled the simplified-fullscreen-ui flag in chrome://flags (i.e. do you see a white popup or black popup when you go into fullscreen mode).

Thanks very much, testers!

Can reproduce on 50.0.2661.94 m windows 7

#31: Fullscreen is still broken with the simplified UI. As for other flags, my command line contains the following switches:

/usr/lib/chromium/chromium --ppapi-flash-path=/usr/lib/PepperFlash/libpepflashplayer.so --ppapi-flash-version=20.0.0.228 --window-depth=24 --window-depth=24 --flag-switches-begin --enable-devtools-experiments --enable-nacl --enable-password-generation --disable-smooth-scrolling --flag-switches-end

A tab created with the "duplicate" functionality does not always have the full screen issue even if the original tab has it.

I've done about as much as I can on this (I don't understand the fullscreen code so I'm not sure how to proceed investigating). Assigning to ananta@ to investigate further. I don't really think your other bug would have caused this but you seem to know something about the fullscreen system on Windows.

Also +scottmg and +jschuh for visibility. This is a potentially bad issue but we haven't got a good repro case yet.

Also having that issue.
It started to happen together with Issue 604359.

Here are my details:
- Operating system version (e.g., Windows 10).
 Windows 10 Version 1511
- Chrome version (e.g., 50.0.2661.94, see chrome://version).
50.0.2661.102
- 32 or 64 bit Chrome (see chrome://version).
64 bit
- Number of monitors (I wonder if dual monitor is related).
3
- Graphics card manufacturer and model (e.g., Nvidia GeForce GTX 970).
Asus STRIX (Nvidia) Geforce GTX 980 Ti
- If you have disabled the simplified-fullscreen-ui flag in chrome://flags (i.e. do you see a white popup or black popup when you go into fullscreen mode).
No. Not disabled. It's set to default.


I would like to try the custom builds of chromium, but the problem with this issue is that it's hard to reproduce. It works for hours (maybe even a day without the issue) and then out of sudden the tab doesn't change to fullscreen anymore. Closing the tab and reopening fixes it.

Thanks for the detailed report. If you can't reproduce it quickly (i.e., it takes hours) then you can't really bisect unfortunately. Another Q: do you typically have lots of tabs open or very few (can you give me a rough estimate, e.g. 2 or 3, tens, 50, 100?)

The fact that it only happens within one tab and after a few hours indicates it's something going wrong in the renderer process, possibly due to running out of memory or something like that.

Yes.
I do have many windows and tabs open.
Around 50 usually and in some cases it can even reach up to 100.

I'm not sure if there is any internal limit but as I do have the 64bit version of Chrome there shouldn't be any "soft" memory limit (like 4GB on 32bit). I do have 32GB of RAM and even so I haven't checked yet, I'm pretty sure it's not at it's limit when this issue happens. Also everything else works as I can watch the video and even going into Fullscreen manually (F11) does work without issues.

#44 Yeah so my understanding is that once you have a certain number of tabs open, Chrome starts coalescing tabs together into the same renderer. You could have a renderer that is getting into a bad state (not necessarily out-of-memory) due to too many tabs. F11 doesn't go through the renderer so it would explain why F11 works but HTML fullscreen doesn't.

Able to replicate on Chrome Stable 50.0.2661.102 m (64-bit). Restarting chrome fixes the problem.

#45 Until a month ago I used to have even more tabs without issues as I kept my computer running for days. Cause of another reason I do restart my computer daily now which kills open tabs/windows. So I'm pretty sure whatever it is, it got introduced with the update to Chrome 50 as it didn't happen before with even more tabs open.

I can consistently reproduce both fullscreen and partially fullscreen. 
Complete fullscreen only happens when I open youtube in a new tab in an existing window.

Partial fullscreen only happens when I open youtube in a new tab and then drag off the tab so that it becomes it's own new window.

Maybe that will help?

Deleted: partial.png 199 KB

	partial.png 199 KB View Download

Deleted: complete.png 195 KB

	complete.png 195 KB View Download

I am also on Windows 10, Chrome Version 50.0.2661.102 m (64-bit).

#49 Thanks for the tab dragging clue. That may be the cause (a lot of people have said it happens "intermittently"; maybe it's happening when they drag off a tab but haven't been able to connect the dots).

I can vouch for the new window tab. That's exactly how it happens to me
usually.

I still can't reproduce this from dragging a YouTube tab off and trying to fullscreen it :(

It could have to be something with tab dragging and pulling out of existing tabs in a specific way as I sometimes do that, too, and it could be the reason when it starts happening for me. But I can't reproduce it by simply dragging an existing tab to create a new window, too. Works as expected for me. Also it happened for me about 2 days ago with only 10 Tabs open (restarted Chrome about half an hour before it happened) so the number of tabs doesn't matter it seems.

#49 if you can reproduce it constantly maybe you can bisect as #37 descibed to figure out what caused it?

I was having this issue as well. i manually performed an update on chrome, and it fixed this problem.  To do this, go to settings>about. It should automatically check for updates and update to 51.0.2704.63 m. Click the relaunch button, and it should fix the bug.

The same bug again occurs and is not fixed .I updated to version 51.0.2704.63 m and again this bug with  windows taskbar shwoing while watching you tube and twitch tv in full screen.It works only if you restart chrome after the bug is appeared and then next time when taskbar apears in full screen same thing restaring and you will fix it or f11 but that doesn't fix the problem permanetly.I tryied to reinstall chrome settigs then reinstaling the whole program but the problem still persists and there is no hope I guess nobody can slove this problem even after updating the graphic drivers problem is still here.Can someone with higer knolege please tell use what is the cause of this  bug if it's not chrome related then what is it ? and when google will realease a stabble bug free build without start menu problem related with windwos xp,vista,7 8,10  ?

When I updated my Mac (OS X 10.11.5) from Chrome 50.0.2661.102 or 50.0.2661.94 (the other entries in /Applications/Google Chrome.app/Contents/Versions) to 51.0.2704.63 (64-bit), this bug started to happen to me. Could it be that the fix people (e.g. comment 55) are reporting in this version fixed the problem on Windows and inadvertently caused the same problem on Mac?

#57: I don't think so. We've had others report the problem with 51.0.2704.x (#13), so I think #55 is just not seeing the bug at the time of writing (it seems to be very flaky). This is the first I've heard of it on Mac though.

ananta@: Have you had a chance to look at this yet?

#57 actually happened to me several times on 50.0.2661 this weekend. It's not 100% and I'm not sure what would cause it, but I'll keep my eye out to see if I notice anything weird happen with it on OSX.

Interesting, #59. For what it's worth, this issue was gone again as of last night. I am still using the same Chrome build on the same system and OS version as described in #57; Chrome has not been restarted since it was relaunched to apply the 51.x update, just a couple hours before I posted #57.

+Pangu and Rachel if you can help us take a look. 

spqchan said (in an email thread):

> It rarely happens, but I was able to replicate this several times on my personal
> machine. I doubt this is an issue with the browser UI. Something interesting I
> found is when the browser fail to enter fullscreen for the video, you can open a
> new tab, navigate to the same video in the new tab and then successfully enter
> video fullscreen.

My hunch is that something is breaking in the renderer process. Based on the fact that a) people say they get it after a long time with many tabs open (perhaps renderers combining many tabs), and b) people say if you open a new tab it fixes the issue for that tab (perhaps a new renderer has started).

This means we can (hopefully) rule out browser-side code as the problem, and just look at the Blink code and renderer-specific code (IPCs, etc). I had a brief look at some CLs in this range (49..50) yesterday but nothing obviously causing this problem.

A few days ago I had this issue happen about 3 hours after restarting the computer and with only about 10-20 Tabs open. :-/
I still see no way to reproduce the problem beside just waiting for it to happen which seems very random.

Have also seen this multiple times over the last week. Version 50.0.2661.102, OS X 10.11.5. Is there a debug mode I can run in that might capture something useful the next time it happens?

If I switch to a different profile, the fullscreen start to work again.

So, I confirm this issue, and I confirm that it can be "worked around" by creating a clean profile.

I have a very similar issue but for me the taskbar is showing then I open a video in fullscreen view. It's happening on Chrome 52 Beta, 42 Stable and 50 Stable. It's also happening to me very, very often. Almost always. This issue started happening to me around April. The same as #comment48 said. I think this my be related to one of the Windows Update patches. I'm also an Optimus user with Intel HD3000 and Nvidia 520M.

Using Asus laptop with 15" HD display on Win7 x64. Chrome version is 52.0.2743.24 beta-m (64-bit).

Deleted: YTfullscreenFktup.png 1022 KB

	YTfullscreenFktup.png 1022 KB View Download

#66: Does that mean it's fixed completly if you create a new profile or just a temporary workaround before it breaks again? Cause at least for me closing the "broken" tab and reopening a new one just fixes it. No need to switch to a new profile.

#67: That's issue 604359 even so it should be fixed already in Beta 52. Also as far as I can see it should have not affected Chrome 42 as it started with 50. Maybe you should give Canary a try to see if it works there?!

Chrome 42.0.2311.135 (64-bit). 
Same thing. For test I used the video linked above: http://www.w3schools.com/html/html5_video.asp

 Issue 618438  has been merged into this issue.

#68: I am not really sure... at the moment I can't reproduce the issue anymore (I didn't upgrade, I just rebooted).

So, maybe the issue can be solved by rebooting.

I am able to reproduce the issue on my personal PC and what I have observed was this is not just the Video not playing in fullscreen but entire full screen logic is broken since when I hit F11 on my PC it doesn't go into full screen mode.

I will get the bisect later tonight. 

#68: I just checked the Canary build and the taskbar is still there.

#72: F11 button also doesn't solve the problem for me. Taskbar is there. What helps for me is clicking at the taskbar and back on the video. It's pretty frustrating though. 

Version 53.0.2756.0 dev (64-bit)
Linux xubuntu-desktop 4.4.0-21-generic #37-Ubuntu SMP Mon Apr 18 18:33:37 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

Just started happening with the latest canary build, but F11 still works.  Have also been plagued with complete system freezes when Chrome Dev is running, not sure if related.  Haven't found anything in the logs

Please find the bisect range which I got on my personal PC :

You are probably looking for a change made after 381928 (known good), but no lat
er than 381935 (first known bad).
CHANGELOG URL:
  https://chromium.googlesource.com/chromium/src/+log/763bbe944c22635b01caaa86cceb65a21fa1bb4c..9cc8ac7b7c2c95836b603afdf9ea6fa57a2b100c

Suspecting Cl : https://chromium.googlesource.com/chromium/src/+/70bdd0c31e09231f9690185313fb0ba2df3dc12a 

I see a few reports of this happening on Linux and Mac, which aren't affected by my patch. But it's possible there are two separate bugs.

#75: Thanks for doing the bisect, pbommana! How confident are you that you've got the right range? (As this issue is quite difficult to reproduce, how confident are you that when you answered "good" in the bisect tool that it was actually good?) Did you bisect on Windows?

#77: I'm not sure what that comment is but it just had two attachments with no file extension and no context. Assuming spam, so deleting.

Well... I finally repro'd this myself on the weekend (on a personal Windows 10 PC). I took some notes and screenshots. Here are some facts I found:

- The "Press Esc to exit" bubble is not shown (I think we already knew this).
- It started happening for me on gaming.youtube.com.
- It affects only one tab (at the time, this tab's renderer process was 160 MB and it was the largest process in my process list). All the other tabs were functioning correctly.
- If you navigate or refresh within the same domain, the bug persists.
- Surprise: If you manually type in a completely different domain into the same tab, the bug persists. I confirmed that the new page was in a new renderer process (new PID). So the bug is tied to the *tab*, not the *renderer*.
- I navigated from gaming.youtube.com to permission.site and the issue persisted. When entering "fullscreen" on permission.site, the entire site contents would slide upwards by a few pixels. Not sure what that means. Attaching screenshots.

Deleted: ytgaming-after-fullscreen-failed.jpg 366 KB

	ytgaming-after-fullscreen-failed.jpg 366 KB View Download

Deleted: ytgaming-task-manager.jpg 390 KB

	ytgaming-task-manager.jpg 390 KB View Download

Deleted: permissionsite-before.png 125 KB

	permissionsite-before.png 125 KB View Download

Deleted: permissionsite-after-fullscreen-failed.png 125 KB

	permissionsite-after-fullscreen-failed.png 125 KB View Download

Still facing the problem. It sucks for months (?) now.
Version 51.0.2704.103 m (64-bit). Windows 7 x64.
Not very much tabs open, mostly 2-3. 

It turns out that this bug is really useful sometimes. For example when a website has a lot of clutter around the player or forces the player to have a certain minimum size. Then I can click the fullscreen button and enjoy the video uncluttered while still being able to do other work because the video only fills the tab body.

So maybe once this bug has been fixed the chromium developers could make it possible to enable this behavior via some kind of option or browser extension.

i have only been able to reproduce it under 2 monitor display, and moving the tab from one monitor to another

Had an idea upon reading #81: Do the separate desktop spaces on Mac OS X work at all like separate displays? When Chrome is fullscreened on Mac its window animates into a new space to the right of the primary desktop. Could be that makes the system behave like it has multiple displays.

That's pure speculation, but if this bug persists on Mac it could be worth exploring. I haven't run into this in a while, personally, but I also have been watching a lot less YouTube the last couple weeks, so there's much less opportunity for it to happen.

Re: #9 I see the bottom half of the Awesome Bar sometimes in full screen, but I have Bookmarks hidden. Could be the same (separate) issue with fullscreen UI hiding.

Hello, i'm getting the same problem, right after the update was installed on this machine. the current version being reported from 'chrome://help' is;
Version 51.0.2704.103 m (64-bit)

Just happened to me with a chain of events nobody has mentioned yet.

1. Be watching a video on YouTube, full screened.
2. Have Google Calendar in a different tab pop a notification, which forcefully removes you from full screen.
3. Navigate back to the YouTube tab in question.
4. Try to re-enter full screen, it fullscreens incorrectly.

I get this as well. Happens at least once a day, generally takes about 3-6 hours to get triggered, then tends to propagate to all of my new tabs very quickly.

Very annoying.


Google Chrome Version:	51.0.2704.103 (Official Build) (64-bit)
Standalone, System install
Windows 7 x64

Two interesting observations to add to what Matt has documented in #78 already:

 (1) Even while `Tab fullscreen` is in the failure state, `Browser fullscreen` (activated by pressing the F11 key) still works.

 (2) The tab brings its failure state along when it's detached and moved to a new window.
 
Obviously I could only reproduce the issue a single time.  Still, after staring at the code for some time, in FWIW-fashion, here's some hand-wavy Saturday-evening mental debugging.

`Tab` and `Browser` fullscreen both end up calling into FullscreenController::ToggleFullscreenModeInternal pretty quickly, and it does not look like the value of the |option| parameter makes much of a difference afterwards.

The checks done before reaching this shared method depend on the states of the ExclusiveAccessContext/-Controller, the FullscreenController, and WebContents::GetCapturerCount(). Except for the last one, all of these are window-specific (i.e. owned by Browser), so I assume that the state is not carried along when the tab is detached. That leaves us with WebContents::GetCapturerCount(), consulted here:

https://cs.chromium.org/chromium/src/chrome/browser/ui/exclusive_access/fullscreen_controller.cc?rcl=1467472681&l=417

The capture count is incremented/decremented manually, which is something that is always suspicious. Plus, there have been some CLs that had to do with tab capture, going in before M50 branch point, so while this is admittedly a long shot, at this stage it's probably worth exploring.

@George, you landed some CLs in this neighborhood. Can you think of any corner cases where we might not get around to call WebContentsImpl::DecrementCapturerCount?

Reading through the comments again, it seems that some of you have experienced the exact opposite. There might be two different issues present here.

The next time anyone encounters the issue, could you please document:
 (1) Whether pressing F11 in the same tab still activates `Browser fullscreen`.
 (2) Whether the failure state is carried along if you detach the tab into a new window. 
 (3) Whether other tabs in the same window can be fullscreened (from the content, i.e. not by pressing F11).

@engedy to answer your questions:

1) f11 actuall full screens correctly
2) doesn't work in detached window
3) other tabs are fine

@engedy
my answers are the same as dhaval

Same. 

Also it seems like duplicating the tab results in a new tab without the issue, though the original tab still has the issue. 

@engedy

same answers as the others, with a bonus about detaching:

* if the broken tab has the video pseudo-fullscreen when you detach it, then in the new window it becomes normal size, and can no longer be fullscreened at all, via the controls or pressing f (and sometimes the bottom right youtube controlls including fullscreen disappear)
* if the broken tab but the video is normal size when you detach it, then you attempt to fullscreen in the new window (hitting the issue), then you drag it back into the original window, it becomes normal size and the full screen state can no longer be changed at all in that tab, even if you detach it again

Thanks for the feedback!

I have explored this avenue a bit more and got lucky with reproducing the issue. I believe I have finally tracked this down.

The proximate cause is indeed the WebContents having a |capturer_count_| > 0, in which case this half-full-screen mode is intended, for the reasons described in [1]. However, the root cause is not something strange in WebRTC, as previously suspected, rather it seems to be a code path taken in the ThumbnailTabHelper that is not calling into [2], thus not decrementing the capture count.

Unless I am not missing some CLs to the opposite effect, the culprit lines have already been refactored away by [3], so the issue should be fixed on Canary/Dev.

@Jayson, @Christopher, could you please confirm? What would be the best way to fix this on the M52 branch?

@George, false alarm, sorry.

[1]: https://cs.chromium.org/chromium/src/chrome/browser/ui/exclusive_access/fullscreen_controller.h?l=37-48

[2]: https://chromium.googlesource.com/chromium/src/+/7fb8a56234863c20dc9d2bce31f730152d086dfc/chrome/browser/thumbnails/thumbnail_tab_helper.cc#223

[3]: https://codereview.chromium.org/2093113003

This is a very big refactor and probably isn't reasonable to push into M52, considering how close that is to stable.

Yes, it is likely that that is the cause, and the work in  issue 623716  will fix it. As of now, the increment/decrement in thumbnail code are gone from TOT.

The only option for M52 is to put a "#if defined(OS_MACOSX)" around the increment/decrement capture counts, since they are only needed for Mac.

This will leave this bug on Mac, but at least will fix other platforms.

What if we moved the increment call to after the two checks in AsyncProcessThumbnail(), i.e. line 171 in [2]?

It is reached synchronously from the current call site, so it looks safe to do; and other than those two checks I think everything makes sure to fire the callback, so it should resolve this issue without introducing a regression on the other.

I would just like to add my opinion to this thread: I LOVE when I get the "Partial/Tab Fullscreen". It allows me see a youtube video at a larger size without having to sacrifice the entire screen (which, on a laptop, is my only screen). I can still move between tabs and programs and do work, but leave the "nearly-fullscreen" video on in the background.

The "tab fullscreen" also avoids the annoying OS X fullscreen animations.

This is how Netflix is all the time - their videos take up 100% width and height of the browser windows - and I think that's a really useful way to present video in a browser.

Since this is an intentional feature in Chrome (allowing "partial fullscreen" under certain circumstances), why not make it an option/preference for the people who like larger videos, but only have a single screen?

WRT #97, I don't feel comfortable doing anything in M52 -- it's too fragile.

Adding M53 merge request before I forget!

WRT #98, you can "enable" this mode by casting to a dummy ChromeCast (using the self-mirroring extension at https://bugs.chromium.org/p/chromium/issues/detail?id=371081#c13). It's not great a great setup for battery, though.

Merge completed to M53 (https://codereview.chromium.org/2156333002)

This merge is not approved yet for M53. Per comment #100, is it already merged to M53?

It is merged.

Still broken... This is my problem...  Look at the top and the left of the picture. That's my wallpaper you can see.

Deleted: 3.JPG 99 KB

	3.JPG 99 KB View Download

@104, that looks like a different issue. Please file a new bug at crbug.com/new

I am having trouble watching any videos through Chrome in full-screen mode. The task bar doesn't get hidden for anything (YouTube, Netflix, etc).

Deleted: Capture.PNG 2.8 MB

	Capture.PNG 2.8 MB View Download

Please file a new bug for this issue.