This sounds reasonable to me but I don't know enough about how we view kiosk mode. +rbyers@ and +mustaq@ who have/are looking at user gesture indicators. Is this something we'd consider relaxing?

Hmm, I would have expected kiosk mode chrome in full-screen w/o a requestFullscreen call!  So my first thought is that this is not a gesture-related problem.

Clarification questions:
- Do we expect multiple windows in the mode?
- Or even multiple tabs?

my problem is not the browser itself in fullscreen mode(which works fine) but one specific dom-node in fullscreen mode, for example a <video> tag.

i'm using a single tab in kiosk mode, the user has no options to navigate between tabs. 

i think this is the more common usecase.

So the problem is that you can't play a fullscreen video w/o a gesture, right?  Then the test page above doesn't apply.

Could you please create a minimal test for this?

mlamouri@: Looks like it's your domain?

Re-assigning to you mustaq@.

Indeed, OP is trying to make <video> fullscreen but this is using the Fullscreen API which requires a user gesture.

Maybe we could use a similar system as the autoplay whitelist from Web APK with kiosk mode? That whitelist could apply to fullscreen?

+foolip@ as it is fullscreen related.

i don't see a different behaviour trying to switch a <div> to fullscreen or a <video> to fullscreen.

both fails when there was no user gesture, both fails with the same error.

i think the attached example is minimal, you want me to do s/div/video/g and attach a new example? 

google@systemstart.net: your repro is fine since it repros.

foolip@: Can we whitelist the Fullscreen API with kiosk mode as Mounir suggested?

I don't know what kiosk mode is, but if Chromium is used to render non-web content that is entirely under the control of the kiosk maintainer, the bypassing the user gesture checks seems fine. Note that this is not really a security check, but a user annoyance mitigation.

Note that if the same configuration is ever used to browser the web, that might be a problem, because there may well be requestFullscreen() calls out there copy-pasted and forgotten that have never worked, that then start going fullscreen where it actually was not intended.

hi.

i take from this discussion that a fix seems legit.

as i am working for a company which would use this feature, a donation 
or bounty would be possible, if the outcome would be a patch or fixed 
version.

cheers,

alexander

On 07.06.2017 15:25, foo… via monorail wrote:

#9 - what about an even simpler logic? If the document is already in full screen mode (be it due to kiosk mode or due to already calling requestFullscreen with a user gesture beforehand), let any call to requestFullscreen pass without a user gesture.

It should not matter to the user whether the page brings an element to the full screen or stays full screen with the entire document and the page can do it in other ways (hiding everything else from the page using JavaScript and put the element forward instead), so you are not preventing anything anyway.

#11, I think something like that could make sense, at least if the element's node document is already fullscreen. (An iframe inside a document that happens to be fullscreen should probably not be exempted.) Then there's the question about how to detect kiosk mode, which I don't know anything about, but it could reasonably behave as if the top-level document is fullscreen.