Re: Comments and requests.

At 07:34 PM 10/19/97 -0400, Alexandre Rafalovitch wrote:
>On Fri, 17 Oct 1997, Mike Champion wrote:
>
>Ok, just to be cristal clear.
>
>If I am writing a browser and it uses DOM to represent the parsed
>document and my parser can "make sense" of partially wrong input, the
>erorrs should be discarded as "non-significant" and would not show up in
>DOM.
>
>This way, if internal structure was written out from the DOM structure,
>any errors that were in original document would not appear in the output?
>

There's no way to be "crystal clear" here because it's not safe to assume
ANYTHING about what a browser that "fixes up" invalid HTML will do to
represent it in the DOM.  The representation of a non-well-formed document
in the DOM is "undefined" at present, so different vendors may do different
things.  If we get lots of feedback indicating that this is unworkable, we
might address it in a later draft of the DOM spec. But the solution is
likely to be more "Stalinist" than "Leninist", e.g., throwing an exception
determining that it is a "non-document" and sending it to the DOM
equivalent of Siberia.

We're not assuming that all documents on the Web will be well-formed by the
time the DOM is released (of course), but we are saying that anyone who
wants to use the DOM to manipulate their Web document had better make sure
that it is well-formed first.

Received on Monday, 20 October 1997 09:51:26 UTC