Re: HTML 2.0 LAST CALL: URI vs URL

Daniel W. Connolly (connolly@beach.w3.org)
Fri, 2 Jun 95 00:34:09 EDT

In message <Pine.SUN.3.90.950601171518.26616B-100000@jobe.shell.portal.com>, Da
vid - Morris writes:
>
>Since complaining about inconsistent usage or URI and URL in earlier
>versions of the draft, Dan has cleaned things up (thats good) and

Section 7, "Hyperlinking" needs more work. I kinda made up my own
definition of "anchor" and I need to go back to using the old
Dexter model definition. I was dazed and confused. Sorry.

>I have read the URI and URL RFCs (thata bad?).

Note that the URL RFC, 1738, is standards track, whereas the "URIs as
used on the web" RFC, 1630, is informational. HTML 2.0 cites 1738
normatively, and 1630 informatively. Hence HTML 2.0 provides its own
definition of the term URI. As a happy conicidence, it matches the
definition in 1630 (at least it will before I'm done -- I got the
fragment identifier stuff wrong.)

>If I missed something, cool, but as I understand the RFCs,
>
>A URL is one kind of scheme of a URI. Hence,
>
>URI: url:http://www.w3.org/somepage.html
>and URL: http://www.23.org/somepage.html

That's not the way I see it, nor TimBL. And since we're the
authors ... :-)

It's more like:

URI: http://www.w3.org/somepage.html
and URL: http://www.23.org/somepage.html

are the same.

The venn diagram in "UR* and The Names and Addresses of WWW objects"
[1] says it best:

_______________________________________________________
| |
| _______________ _______________ |
| | ftp: | | urn: | |
| | gopher: | | fpi: ? | |
| | http: | | | |
| | etc | | | |
| |_______________| |_______________| |
| URLs URNs |
|_______________________________________________________|
URIs

[1] http://www.w3.org/hypertext/WWW/Addressing/Addressing.html
Wed Mar 22 23:30:03 1995

>are equivalent. The distinction being that the URI includes URL:
>to specify the scheme.

URL: is something that the URL RFC suggests that you use, say, in a
plain text document or magazine to let somebody know that this thingy
is a URL, and not line noise.

>Based on my check of netscape (X & Win), Mosaic (win) and
>websurfer, I must concluded that specification of a URI as the
>target of HREF on an anchor is not current practice.
>
>Have I missed something?

I think so. A URL is a URI. The converse is not necessarily true.

>Is there a compelling reason to specify URI and not URL in
>this version of the standard.

The term URI includes the notions of relative addresses and fragment
identifiers, whereas the URL spec doesn't include these. I should read
Roy Fielding's [RELURL] draft again to see if it covers these
sufficiently. Is it an RFC yet? Can I cite it normatively, if it turns
out to fit the bill?

Dan