Re: URL-Reference / "empty URL" question

[Foteos Macrides <[e-mail]>]

Larry Masinter <[e-mail]> wrote:
>Don't "#xx" fragment references work in the result of a POST?

	No UA, to my knowledge, has problems with fragments in the
replies from POST submissions, nor in the replies from submissions
with METHOD=GET, nor in the case of CGI scripts set up to return
the form when the Request-URI has no ?searchpart or POST content.

	The only variations among UAs is in whether they resolve
"lone fragments" versus "(current document)" or "base", and that's
orthogonal to the issue of whether one is dealing with forms.

	There is, of course, the difference that both
URL#fragment				and
Request-URI?searchpart#fragment		work in any context,
	whereas
Request-URI(+ POST content)#fragment	presently works only when
Request-URI(+ POST content)		is the "(current document)".

and nothing can be done about that until real consensus is reached
on how to represent "(+ POST content)", e.g., in bookmark links,
as well as in other contexts.


>I would expect fragment references to work in emailed
>documents, too, even though there is may be no useful
>BASE for the resolution of other relative URLs.

	One can include a <BASE HREF="scheme://host/path"> tag in the
mailed markup.  Some browsers (e.g., Netscape, Lynx) add a BASE tag
to the top of emailed text/html.  That's just a "works, for now"
hack, and they send a Content-Base header as well, which hopefully,
someday, will actually do the job. (so the "works, for now" hack can
be eliminated :)  The URL draft's "lone fragment" rule (with Roy's
"clarification of intent") works fine in that context.

				Fote

=========================================================================
 Foteos Macrides            Worcester Foundation for Biomedical Research
 [e-mail]         222 Maple Avenue, Shrewsbury, MA 01545
=========================================================================