Re: minor fix for url-syntax-02

[Foteos Macrides <[e-mail]>]

"Roy T. Fielding" <[e-mail]> wrote:
>This is from Gisle Aas <[e-mail]>no>.  The regular expression
>in Appendix B can be slightly simplified from
>
>      ^(([^/?#]+):)?(//([^/?#]*))?([^?#]*)?(\?([^#]*))?(#(.*))?
>                                          ^
>                                          redundant
>to
>      ^(([^/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?

	The 02 draft specifies that a lone fragment (HREF="#fragment")
should be resolved versus the current document's URL, rather than
versus a BASE tag's HREF, if one was present.  That's good, because
in most cases you otherwise would be getting another copy of a document
that's the same as what you've already copied locally, to go to
some place in it that's the same as in your current document.  It also
doesn't matter that no presently deployed browser behaves that way, and
Lynx users generally would prefer that behavior, so I went ahead and
made that change in Lynx.

	The 02 draft also now specifies that an empty reference should
be treated as the current document's URL, rather that using the BASE.
That's protentially a problem for FORMs, in which an ACTION might
be omitted with the expection that the BASE will be used, to go
back to a CGI script which generated the current document, and I'm
worried about making this change in Lynx when no other browser behaves
that way.  Could the rationale for this change be spelled out?  Do the
major browsers plan to make this change?

			 	Fote

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