Re: ERB: decision and conundrum

At 11:17 AM -0800 3/17/97, Tim Bray wrote:
>Terry Allen wrote:
>>| >3. The '#' may be followed by the string "<tei>"
>>The angle brackets are excluded characters per
>>    ftp://ds.internic.net/internet-drafts/draft-fielding-url-syntax-03.txt
>>of December 1996, the latest URL syntax draft I can find at the site.

This is eay to fix, though. An alternate syntax should not be a serious

>Not only that; that draft also makes it crystal clear that the URL
>syntax hardwires in the processing model; i.e. if you say URL#fragment,
>the user agent has to get the whole URL and pick out the fragment.
>Clearly this is a nonstarter for XML, where the fragment might select
>an entry from the OED... on the other hand using URL?fragment has its
>own set of problems.  In fact, what we *really* want is to specify the
>URL and fragment by way of policy, and let the implementation figure
>out how to divvy the load between client and server.  Also, we'd like
>to do this without egregiously violating the ruling RFC's.  Which may
>cost us our ability to declare that "locators are URLs".  Blecch.
>Double blecch. - Tim

   No, we need to define the client-side method. We can suggest a
server-side sytnax, but frankly I doubt that we can do that in a way that
will satisfy all the servers (and unprincipled hacks) out there on the
server side -- and I don't see that we really need to.

   I think that leaving it to the author to decide how the URL should be
done is perfectly reasonable: Things like the OED will not be served up
monolithically (probably by contract with OUP!), and so references made by
authors (or people browsing a site) will automatically be in terms of URLs
that designate smaller, well-formed document fragments.

   I'm afraid that I don't see what the problem is with having two formats,
and no standard for the server side.

  Servers can serve up URLs that take advantage of whatever queries they
support. This does not seem difficult to ensure. I don't really see a
significant practical advantage (other than neatness) to standardizing the
server query syntax.

   I suppose you could do blind queries to a server if you have a common
syntax, but I'm not sure that that is a primary goal here...  It certainly
doesn't seem _required_ to support linking and browsing, which are the
de-facto requirements for WWW compatibility.

   Can anyone offer an argument to convince me?

  -- David

David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________