Bug 76478 – REGRESSION: iframes with composited contents are not transparent in WK1

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Simon Fraser (smfr)]

The pixel result of compositing/iframes/resources/repaint-after-losing-scrollbars-iframe.html from DRT shows that the iframes are showing up as opaque white. This reproduces in a WK1 window in Safari.

[Simon Fraser (smfr)]

I think http://trac.webkit.org/changeset/92874 may have broken this. FrameView::useSlowRepaints() returns 'false' for iframes with composited content, but Widget::paint() (in WidgetMac) relies on [[scrollView contentView] copiesOnScroll] to decide when to turn off background drawing.

[Radar WebKit Bug Importer]

<rdar://problem/10710494>

[Simon Fraser (smfr)]

WidgetMac really needs to get at FrameView's m_contentIsOpaque bit, but that would be a layering violation.

[Filip Pizlo]

The relevant test appears to pass as of r115915

[Simon Fraser (smfr)]

compositing/iframes/resources/repaint-after-losing-scrollbars-iframe.html is the iframe source, not the actual test. I probably meant compositing/iframes/repaint-after-losing-scrollbars.html.

[Filip Pizlo]

(In reply to comment #5)
> compositing/iframes/resources/repaint-after-losing-scrollbars-iframe.html is the iframe source, not the actual test. I probably meant compositing/iframes/repaint-after-losing-scrollbars.html.

I figured that's what you meant, and that is indeed what I'm referring to.

[Simon Fraser (smfr)]

This still fails. enne, can you look at this?

[Adrienne Walker]

(In reply to comment #7)
> This still fails. enne, can you look at this?

Sure, I can take a look.

[Adrienne Walker]

Created attachment 140636 [details]
Patch

[Adrienne Walker]

(In reply to comment #9)
> Created an attachment (id=140636) [details]
> Patch

I am not happy with this, but am unsure about how else to handle this.

I don't think it's correct to assume that !ScrollView::canBlitOnScroll() implies !FrameView::m_contentIsOpaque.  It's also very strange to me that drawsBackground is true for something that shouldn't draw a background.  That paint function is a pile of assumptions.

[Adrienne Walker]

smfr: ping

[Adrienne Walker]

(In reply to comment #11)
> smfr: ping

smfr: re-ping?

You asked me to fix this three weeks ago, so I uploaded a patch.  Was this something you wanted landed?  Otherwise, maybe this issue should just be WONTFIXed.

[Simon Fraser (smfr)]

Comment on attachment 140636 [details]
Patch

Sorry for the delay.

[WebKit Review Bot]

Comment on attachment 140636 [details]
Patch

Clearing flags on attachment: 140636

Committed r118840: <http://trac.webkit.org/changeset/118840>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.