XML incremental rendering, was Re: Standards mode and Quirks mode (was Re: [CSS21] Test Suite)

From: "Robin Berjon":
>   .... * therefore, there is not a single argument against incremental
>     rendering of XML documents...

Just only one: document cannot be rendered (drawn) until XML parser is not
sure that
the xml is well formed.

Illustration:

Let's say we have a printer (as an output media) and  following xhtml:

<html>
<body>
<p>paragraph</p>
</html>

When UA (optimistic one) have got the </p> it starts rendering it - draws
'paragraph' on
physical paper page. Then after reading </html> it discovers that document
is not well formed so it is not a xthml at all.
So UA should return paper page back to the printer tray and clear it
somehow? Not bad, eh?!

Even in case of screen media: document appears on screen and then it will
dissolve somehow?
How it supposed to look like actually? If UA will left partial content on
the screen then it means that UA uses input language other then XML.

xhtml recommendation states clearly : Document *must be* well formed.
And there is no such entity as "well formed fragment" in XML.
For any given  XML you cannot render/draw (change state of the view)
partialy.

Or forget about XML.

Andrew Fedoniouk.
Terra Informatica Software, Inc.
http://terrainformatica.com


From: "Robin Berjon":
>
> Brian Sexton wrote:
> >>Robin Berjon wrote:
> >>There is a requirement that the document be in error when it is
> >>discovered that it is no well-formed.
> >
> > On the contrary, if a document that purports to be XML is not
well-formed,
> > it does not meet the requirements of the XML specification and thus, is
in
> > error.
>
> "On the contrary"? We could go over the entirety of epistemology from
> the scholastics to quantum physics to try and figure out whether a
> document is in error before it has been observed to do so, but that
> wouldn't likely get us very far. The fact is, an XML processor
> *processes* XML until it finds it not to be well-formed. I don't see how
> it could guess that it's in error before parsing it to that error.
>
> >>Robin Berjon wrote:
> >>There is no requirement that that has to be known before
> >>rendering can start to work.
> >
> > Perhaps not, but if rendering begins before checking for well-formedness
> > then well-formedness is not checked before rendering begins.
>
> That is correct.
>
> >  If A happens
> > before B then B does not happen before A.
>
> That is correct.
>
> >  This seems like a very simple
> > point; I do not understand why you are contesting it unless, as I
suspect,
> > you simply misunderstood it.  Or perhaps I did.
>
> It is simple to the point that you can immediately reduce it to A !=
> non-A, which is a tautology. I don't see how that advances the argument
> though.
>
> The simple fact is:
>
>   * an XML processor processes a document until it is found in error, or
>     the document ends, whichever comes first;
>   * since rendering is part of the processing being performed on the XML
>     document, there is nothing keeping it from occuring while the
>     document is not discovered to be in error;
>   * therefore, there is not a single argument against incremental
>     rendering of XML documents. In fact, it is even part of SVG:
>
>       http://www.w3.org/TR/SVG12/#progressive
>       http://www.w3.org/TR/SVG12/#multipage
>
> What happens when you document is found to be in error (wiping out the
> entire display, or leaving it there but stopping everything else and
> never getting a load event) is up to the given specification to define.
>
> > No, but making remarks like that, you are being unnecessarily and
> > unentertainingly rude--exhibiting precisely the kind of behavior that
makes
> > public discussion lists and forums more annoying and less productive
than
> > they should be.  Let's keep this list productive, shall we?
>
> There are many ways of being rude; I was merely responding to your
> assumption that you know and understand everything, and that tautologies
> need be explained to others.
>
> -- 
> Robin Berjon
>

Received on Tuesday, 27 July 2004 13:49:48 UTC