There's a style element with a title in the above page. You're treating it as alternate. It should, I think, either be preferred or persistent (the spec is vague). I see no reason that it should not be enabled when the page loads. If this works, the above page, http://www.fas.harvard.edu/~dbaron/css/fonts/aspect_dom , will act just like the page http://www.fas.harvard.edu/~dbaron/css/fonts/aspect .

This is semi-intentional behaviour. Peter and I are currently discussing it.

The intent here is for that sheet to become preferred. The content sink isn't doing that right for style element (I have it fixed, it'll be checked in right after M6). The outstanding question is: is the title attribute of the style element significant? Right now, the code treats it just like the title on a link element, ie: it sets the title of the stylesheet. Therefore, any sheet with a title is either preferred or alternate (depending on the title and the possibly previously set default style). The HTML spec is silent on this issue. It explicitly mentions the title attribute of the style element, but doesn't say whether or not it sets the sheet title or just the element's title. Letting it set the sheet title enables the only current way of creating alternate or preferred inline sheets.

Yes, the title attribute of the style element is significant. The spec says as much about the LINK element's title in the section on the LINK element as it does about the STYLE element's title in the section on the STYLE element. Both attributes are tagged as "advisory title". In another chapter, the spec uses LINK as an _example_ of how to link stylesheets. It doesn't say that it is the only way to make persistent/ preferred/alternate sheets. It mentions the "title" attribute, but it does not say that _only_ the LINK element's title attribute will do. The "title" attribute on LINK currently has exactly three uses: 1. Displaying as a tooltip in the unlikely case of someone like me doing a display:block with generated content, 2. Getting around bugs in DOM implementations, 3. Setting persistent/preferred/alternate stylesheets. Since STYLE's title attribute is documented in exactly the same way, why exclude option 3 from its roles?

This is currently blowing up in reflow code during load. But the title attribute is fixed.

The problem is that the root frame code was never changed to handle the StyleChanged reflow command, and so we're ending up incorrectly processing the reflow command and that's why we hit the assert in the absolute containing block code

Changed Reflow() to handle reflow commands of type StyleChanged

Verified fixed.

I think this is a bug. see http://www.w3.org/TR/html401/struct/global.html#h-7.4.3 The title attribute has an additional role _when used with the LINK element_ to designate an _external_ style sheet. Please consult the section on links and style sheets for details.

Re comment 8: I agree with you, but the Style System drivers do not. See bug 188131 for details.