Bug 93508 – Make the Silverlight CAOpenGLLayer opaque if we know the plug-in contents is opaque to reduce blending

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Anders Carlsson]

Make the Silverlight CAOpenGLLayer opaque if we know the plug-in contents is opaque to reduce blending

[Anders Carlsson]

Created attachment 157273 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 157273 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=157273&action=review

> Source/WebKit2/Shared/Plugins/PluginQuirks.h:48
> +        // NPPVpluginTransparentBool, so we'll always force it unless the plug-in has a "background"
> +        // attribute that specifies a opaque color.
> +        MakeOpaqueUnlessTransparentSilverlightBackgroundAttributeExists,

It might be worth noting in this comment that the docs say "a null value for the Background property of the Silverlight plug-in is interpreted as the named color White.", so no colorValue param means opaque.

> Source/WebKit2/WebProcess/Plugins/Netscape/NetscapePlugin.cpp:557
> +    if (backgroundValue == "#000000")
> +        return false;

This is fragile, as it would break if the content changed. Sucks that these aren't CSS colors :(

[Anders Carlsson]

(In reply to comment #2)
> (From update of attachment 157273 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=157273&action=review
> 
> > Source/WebKit2/Shared/Plugins/PluginQuirks.h:48
> > +        // NPPVpluginTransparentBool, so we'll always force it unless the plug-in has a "background"
> > +        // attribute that specifies a opaque color.
> > +        MakeOpaqueUnlessTransparentSilverlightBackgroundAttributeExists,
> 
> It might be worth noting in this comment that the docs say "a null value for the Background property of the Silverlight plug-in is interpreted as the named color White.", so no colorValue param means opaque.
> 

Good idea. Fixed!

> > Source/WebKit2/WebProcess/Plugins/Netscape/NetscapePlugin.cpp:557
> > +    if (backgroundValue == "#000000")
> > +        return false;
> 
> This is fragile, as it would break if the content changed. Sucks that these aren't CSS colors :(

I agree. I'll do the more extensive color parsing in a followup patch.

Thanks for the review!

[Anders Carlsson]

Committed r125081: <http://trac.webkit.org/changeset/125081>