W3C home > Mailing lists > Public > www-qa@w3.org > September 2003

Re: QA Tips: Make readable URIs

From: Dominique HazaŽl-Massieux <dom@w3.org>
Date: 18 Sep 2003 17:45:49 +0200
To: Victor Engmark <engmark@orakel.ntnu.no>
Cc: www-qa@w3.org
Message-Id: <1063899950.19335.261.camel@stratustier>
Le lun 15/09/2003 ŗ 15:55, Victor Engmark a ťcrit :
> Attached as instructed on the web page.
> I sent this a few weeks ago, but it doesn't seem to have reached you...

Sorry for the bad turn-around... I have posted your document as draft
tip on the tips list:

This opens the 2 weeks period for review... I do have some comments on
the content of the tip:
- one of the principle behind the design of the URIs is that they are
opaque, which means that nothing/nobody should infer anything from the
characters used in the URI; this should be reminded at the very top of
the tip, and maybe the title of the tip should be changed to make that
- it's probably a good idea to make your URIs readable, for reasons
close the ones you give; ie, when people have to type URIs instead of
following links, it's much easier to type "readable URIs" than very
complex ones; still, I don't think the tip should speak about
"remembering URIs" for casual users, nor the bookmark use case seems
very relevant. I would say something like
"When a URI has to be advertized through a medium that doesn't easily
support following hypertext links (e.g. a URI on paper or in some email
clients), it's much easier for the user to have to type a readable URIs
rather than a unreadable one."

I simply don't agree on the recommendation of using directories
generally speaking. The organisation of a Web site is really too
dependent on the server configuration, the information architecture, the
publication system to make such a general recommendation. I think it is
a good idea to avoid using query strings as a way to indicate separate
content... The CHIPs note says more succintly:
"CP 1.1      Use short URIs as much as possible "
"CP 3.1     Serve dynamic content with technology-independent URIs"
which probably conveys some of the ideas behind this fact.

In your tip, you say:
"Most (all?) web servers will serve ..."
You can strike the 'all?' parenthesis; this depends on the configuration
of the server in most cases, so it's certain that it's not on all
servers. Regarding the use of shorter URIs, content negotiation has the
same effect as the use of directory, with even one less character; this
should be the proposed solution.

I like the part about language negotiation, although it would be nice to
have this notion appear in the text.

I don't like the idea of localizing URIs; it goes way too far in the
readability of URIs and against the opacity of URIs. Typing URIs doesn't
happen that often that you would have to maintain several structures for
your URI space. Besides, the proposed solution is too specific (symbolic
links doesn't exist on all OS); the most generic solution would probably
involve using HTTP redirects.

In the reference section, the link to your web page should be removed
since it's not about the particular topic of the document (you can of
course link it from the <address> of the tip), and further documentation
on Content/Language Negotiation would be interesting (ie, not only the
way to do it in Apache, but the general specification).

- you should use example.org URIs rather than site.com (the example.org
domain has been instantiated for this exact purpose)

To summarize, I think that the tip addresses a valid question, but would
benefit from a bit more work, and maybe from being based on the CHIPs
note which addresses severeal of this problem. Is that something you
would be willing to do?

Thanks for your contribution!

Dominique HazaŽl-Massieux - http://www.w3.org/People/Dom/

Received on Thursday, 18 September 2003 11:45:51 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:40:34 UTC