Bug 100249 – Add setMediaTypeOverride to window.internals.settings

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[Mikhail Pozdnyakov]

WebKitTestRunner needs testRunner.setMediaType as it's used in fast/media/print-restores-previous-mediatype.html.

[Kenneth Rohde Christiansen]

It is better to convert it to a window.internals.settings

like setMediaTypeOverride, that also makes it cross platform

[Mikhail Pozdnyakov]

Created attachment 170590 [details]
patch

[Kenneth Rohde Christiansen]

Comment on attachment 170590 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=170590&action=review

> Source/WebCore/page/Settings.cpp:128
> +    , m_mediaTypeOverride("screen")

why not pick the current one? Some ports might change it.

[Kenneth Rohde Christiansen]

Comment on attachment 170590 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=170590&action=review

>> Source/WebCore/page/Settings.cpp:128
>> +    , m_mediaTypeOverride("screen")
> 
> why not pick the current one? Some ports might change it.

or say someone does this while printing? then it will be 'print'

> Source/WebCore/page/Settings.cpp:484
> +    mainFrame->document()->styleResolverChanged(RecalcStyleImmediately);
> +    view->layout();

I think the setDeviceScaleFactor does this slightly different

[Build Bot]

Comment on attachment 170590 [details]
patch

Attachment 170590 [details] did not pass win-ews (win):
Output: http://queues.webkit.org/results/14569516

[kov's GTK+ EWS bot]

Comment on attachment 170590 [details]
patch

Attachment 170590 [details] did not pass gtk-ews (gtk):
Output: http://queues.webkit.org/results/14559585

[Kenneth Rohde Christiansen]

Comment on attachment 170590 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=170590&action=review

> LayoutTests/fast/media/print-restores-previous-mediatype.html:8
>      if (window.testRunner) {

shouldnt we check for internals as well?

>> Source/WebCore/page/Settings.cpp:484
>> +    view->layout();
> 
> I think the setDeviceScaleFactor does this slightly different

setNeedsRecalcStyleInAllFrames();
m_page->pageCache()->markPagesForFullStyleRecalc(m_page);

This is what setDeviceScaleFactor does, I am not sure of the exact difference

[Kenneth Rohde Christiansen]

> setNeedsRecalcStyleInAllFrames();

> m_page->pageCache()->markPagesForFullStyleRecalc(m_page);

I think this boils down to whether this is supposed to affect globally (ie. change things in cache and everything alive), or just the page.

What does other settings do?

[Mikhail Pozdnyakov]

(In reply to comment #8)
> > setNeedsRecalcStyleInAllFrames();
> 
> > m_page->pageCache()->markPagesForFullStyleRecalc(m_page);
> 
> I think this boils down to whether this is supposed to affect globally (ie. change things in cache and everything alive), or just the page.
> 
> What does other settings do?

other settings live with just with setNeedsRecalcStyleInAllFrames, I do not think this should be applied to all pages

[Mikhail Pozdnyakov]

Created attachment 170620 [details]
patch v2

Changed the style update in Settings::setMediaTypeOverride, hopefully fixed build on MAC, WIN and GTK bots.

[Kenneth Rohde Christiansen]

Comment on attachment 170620 [details]
patch v2

why didnt you remove the dead qt code? isnt it dead?

[Mikhail Pozdnyakov]

(In reply to comment #11)
> (From update of attachment 170620 [details])
> why didnt you remove the dead qt code? isnt it dead?

should be dead. would do it in a separate patch when this one lands..

[kov's GTK+ EWS bot]

Comment on attachment 170620 [details]
patch v2

Attachment 170620 [details] did not pass gtk-ews (gtk):
Output: http://queues.webkit.org/results/14546619

[Mikhail Pozdnyakov]

Created attachment 170625 [details]
patch v3

another attempt, added missing ';' to symbols.filter

[WebKit Review Bot]

Comment on attachment 170625 [details]
patch v3

Clearing flags on attachment: 170625

Committed r132480: <http://trac.webkit.org/changeset/132480>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.