W3C home > Mailing lists > Public > public-ldp@w3.org > June 2014

Re: best practices comments

From: John Arwe <johnarwe@us.ibm.com>
Date: Mon, 9 Jun 2014 10:40:06 -0400
To: public-ldp <public-ldp@w3.org>
Message-ID: <OF1D91089F.587662FD-ON85257CF2.004E1F6D-85257CF2.0050943E@us.ibm.com>
Having been an "interested party" in 3986 due to previous W3C work, I can 
say second-hand that the authors were being quite precise in their use of 
terms.
Normalization as used in 3986 section 6 is different from what the LDP 
Submission calls a canonical URL.

The intent behind 2.11 canonical URLs is larger/wider than what 3986 
section 6 means be normalization (which is always "for a purpose", and in 
2.11 I'm not sure what I'd call the purpose).  It is intended to handle 
problems like links that come from both sides of a reverse proxy.  E.g. if 
I have a resource http://example.org/123 that is _also_ known (inside some 
firewall) as http://rtc-5.ibm.com/123 , the Submission's intent was that 
the server (the only entity capable of asserting with authority that the 
two URIs identify the same resource) would return a URI (called the 
'canonical' URI by the Submission, a term of our own invention not based 
on any existing source) that all clients could use to know if the 
resources are actually the same.

As we just discussed on the call, the Submission made no assertion that 
its canonical URL would by accessible by all clients - this might be an 
issue in the firewall case above, as well as the http: vs https: example 
Cody cited from the Submission (if a resource is only accessible in 
practice while authenticated, its canonical URL might still be http:).


Best Regards, John

Voice US 845-435-9470  BluePages
Cloud and Smarter Infrastructure OSLC Lead


Cody Burleson <cody.burleson@base22.com> wrote on 06/09/2014 08:42:57 AM:

> Now, the funny thing is, I cannot for the life of me find any 
> original definition of the term "canonical URL". From what I could 
> find, it seems that "canonicalizing" a URL is the process of 
> "normalizing" the URL, which is described here:
> 
> http://tools.ietf.org/html/rfc3986#page-38
> 
> With a little verification that I am correct, perhaps we should say 
> something like:
> 
> URL canonicalization (URL normalization) is the process by which URLs
>  are modified and standardized in a consistent manner; see section 
> 6, Normalization and comparison in [RFC3986].
Received on Monday, 9 June 2014 14:41:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:03:12 UTC