Re: High-Quality Dynamic SVG Viewer

"Chris Lilley" <> wrote in message
> On Monday, January 20, 2003, 12:09:00 PM, Jim wrote:
> JL> Without the closing tags of the svg, or g, we have enough information
> JL> draw the rectangle, however because we don't know at this stage if the
> JL> document is well-formed,
> well, the evidence at that point is that it is well formed.
> Anglo-saxon law, innocent until proven guilty. As soon as you get a
> </svg> or a </rect> or or the pipe closes or something else that
> indicates its not well formed, then its not well formed.

That's fine, but remember the requirement on rendering is that the render
stops rendering the document on the element _before_ the element that is not
well-formed, therefore whilst we can assume the content is well-formed, as
soon as it turns out not to have been, our rendering of its children was
incorrect, hence the need to change what happens when a document is in

> JL> we cannot render anything according to the current
> JL> spec.  (this wouldn't be a problem of course if we built on SGML,
> JL> than XML, then we could've had closing tags as optional...)
> You seem to be mixing up the static and dynamic views of the document.
> SGML speaks of the static view - you get the whole document and look
> at it, and in SGML you can say the closing tags are optional.

Certainly, but my desire for optional tags wasn't because of this, it was to
correct for the error whereby we discover the root element is not well
formed upon the downloading of the last 5 bytes.  SVG 1.0 and 1.1 says "The
document shall be rendered up to, but not including, the first element which
has an error."  ( )
and in our case we either could not have rendered all the previous elements,
or would have to make them disappear, neither are things which a user agent
would be popular for.  If closing tags were optional, we'd not have this


Received on Tuesday, 21 January 2003 12:44:45 UTC