- From: Paul Mitchell <paul@paul-mitchell.me.uk>
- Date: Mon, 13 Feb 2006 14:04:01 +0000
- To: "www-style@w3.org" <www-style@w3.org>
Anne van Kesteren wrote: > > On Sat, 11 Feb 2006 21:51:28 +0100, Paul Mitchell > <paul@paul-mitchell.me.uk> wrote: > >>>>> As for <style>, supposedly this works: >>>>> >>>>> <?xml-stylesheet href="#xxx" type="text/css"?> >>>>> <root> >>>>> <foo xml:id="xxx"> >>>>> bar { background: green; } >>>>> </foo> >>>>> <bar>test</bar> >>>>> </root> >>>> >> >> In your example, the UA doesn't need style #foo until element id foo >> is encountered, which comes after the style has been defined, not >> before as in the previous example, where the style for element root >> is needed before the stylesheet that defines it has materialised. > > > Not really. Rendering usually starts after parsing. I can see that it > might give some problems for incremental parsers, but those problems > would not be much different from a script inserting a style block > when the parser is halfway... > I'm confused - perhaps we are using different words for the same thing. "Rendering usually starts after parsing". Which common UAs exhibit that behavour? If my connection is a POTS modem and I request a 100MB document, must I wait for the download to complete before I see anything? I would consider that a serious flaw. I'm prepared to wait for the stylesheet to fully load before rendering begins, but no longer. I stand by my original point, which is that the original example (xml-stylesheet processing instruction referring to a fragment of the current document) is incorrect. It is incorrect because it _lies_ to the UA, asserting the existence of a stylesheet that cannot be proven to exist, something a rational UA will not tolerate. -- Paul Mitchell www.paul-mitchell.me.uk
Received on Monday, 13 February 2006 14:05:44 UTC