Re: relation of ISMAP to URL

Dave Kristol (dmk@allegra.att.com)
Fri, 14 Apr 95 09:34:31 EDT

I said:
> > I don't think the HTML 2.0 (or 3.0?) specification states how a
> > browser should modify the anchor URL surrounding an ISMAP image when it
> > sends the augmented URL to a server. That behavior seems to fall into
> > a gray area between HTML and HTTP.
>
"Roy T. Fielding" <fielding@avron.ICS.UCI.EDU> said:
> Not really -- it is an HTML user agent issue as to how to construct the
> ISMAP URL such that it remains a valid URI. My Relative URL draft
> won't help here either, since none of these are relative (and ISMAP does
> not apply to the document's base URL).
Oddly enough, I accepted Larry Masinter's remark that the Relative URL
draft explained it. My interpretation was the following:
1) An ISMAP adds a relative reference ?x,y to a base URL.
2) The base URL is the one in the surrounding <A> anchor.
3) Thus, if there was a prior query string, the ISMAP ?x,y overrides
the prior query string.
>
> I agree that this needs to be specified in the HTML 2.0 draft. I recommend
> changing the ISMAP description of Section 7.3 (Image) as follows:
[nice description deleted]

I applaud the added words, particularly the specification of what happens
when a point in an ISMAP is selected. Thanks for nailing down the units
of x,y (although it's too bad it wasn't, say, a value between 0 and 1 to
indicate what fraction of full-scale x or y the selection was at).

Since current practice is inconsistent on what happens if the base URL
of an ISMAP anchor contains a query string or '#' character, I guess
it's okay to prohibit such things in the base URL for HTML 2.0. But I
think Larry was right that the Relative URL document adequately
describes what should happen. Perhaps the HTML 3.0 document should
simply say that the coordinates are added as ?x,y to the base URL
according to the rules of relative URLs.

Dave Kristol