Re: Implementing NodeList

At 02:29 PM 8/19/98 -0400, keshlam@us.ibm.com wrote:
>I'm confused. I thought we left it at "All NodeLists and NamedNodeMaps
>_are_ live accessors."
>
>The PR is unfortunately no more clear than the WD was in this regard. Some
>instances of NodeList carry words saying that they're live. Others don't,
>but do not say that they _aren't_ required to be. And the description of
>NodeList itself doesn't even mention the issue.

I think the confusion is due to earlier disagreements within the WG and
this mailing list; I'm pretty sure (I'll eat my words if I'm wrong) that
the PR says that all NodeLists *are* live.  John Cowan quoted something to
the contrary earlier that was written *before* this was nailed down by the
WG.  

So, the intent of the WG is that ALL NODELISTS ARE LIVE.  In other words,
all NodeLists should point at nodes in the document, not at static copies,
making the "liveness" manageable for implementers. [Think "lazy evaluation"
for getElementsByTagName() ].  If this is still really confusing in the PR,
please point out any language that implies something different, and we'll
be sure to fix it during the comment/voting period (roughly the next four
weeks) for the actual standard (assuming the W3C membership approves the PR).

So, I'm sorry that there is still ambiguity about this, and we *will* fix
it soon, but I'd appreciate specific pointers to the ambiguous wording.

Thanks,

Mike Champion

Received on Wednesday, 19 August 1998 14:48:40 UTC