Re: Document cleanup suggestions

More stuff from the sanity-check pass (as long as I'm swapping it all into
wetware anyway...)

Text: Currently says that it's the non-markup content of an element, and
that markup will be siblings on either side of it. Two problems with that
statement.

(1) It may be the content  of other things, eg Attribute.

(2) It may be only one contiguous chunk of that content. There may be more
than one text child separated by markup:
     As in <strong>this</strong> case,
where the siblings are Text, Element, and Text again. Current wording
suggests only a single Text should be expected except during editing; not
what you intended.

Has joinText's behavior been nailed down since the 20th? Somehow, having it
return an orphan Text doesn't seem to save a lot of work versus cloning the
first and appending the data from the second.


ProcessingInstruction: You might want to make more explicit that the target
re-appears as part of the data... and remind folks about the PI's syntax,
eg whether or not whitespace is allowed between <? and the target. (I'm
somewhat surprised that you offer target alone, and all-data, but _not_
data-following-the-target. It would seem appropriate, if you're going to go
to the trouble of parsing out the target, to offer both results to the
user.)


DocumentType: Should this be self-initialized to include standard entities
such as %lt;? Or is it assumed that whoever creates the Document will take
responsibility for setting this up correctly? (The latter is more general,
but since these extended classes are XML-specific...?) Might be worth
pointing out once again that the DOM explicitly doesn't discuss creation of
the root Document, and only operates on it once it exists, just to avoid
questions about who shaves the barber.

Received on Thursday, 30 July 1998 16:12:51 UTC