W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1996

Re: URI Pattern Syntax

From: <hallam@w3.org>
Date: Fri, 23 Feb 96 19:36:58 -0500
Message-Id: <9602240036.AA15202@zorch.w3.org>
To: Mike Meyer <mwm@contessa.phone.net>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Cc: hallam@w3.org

>Now, could someone explain what application this would be used for?
>Are servers expected to create multipart objects when handed a URL?
>Browsers expected to ask for a bunch in a row? A shorthand for mirror
>systems?

Yes eventually, Yes eventually, Yes! 

Also, assertion exchange, annotations, server configuration...


I think there are any number of occassions where refering to sets of URIs will 
be very usefull. Some times there will be more work that is justified than 
others. I beleive that regardless of the application many of these will end up 
being entered either by hand or generated with drag 'n drop type tools. Thus I 
favour the structural over the syntactic approach.

Note one trick that VMS allows and UNIX fails on :-

rename *.jpeg *.jpg

The value from wildcard #1 is used as the replacement key in wildcard #2. This 
trick does not extend to well, multiple *s become difficult. But it is a very 
usefull facility that when I ask UNIX appologists how to do it they tend to 1) 
try to explain why I should not want to be able to do that 2) extol the virtues 
of full regular expressions, or 3) unearth a 2453 line perl script which reads a 
directory then produces a csh script which performs the renaming operation.

Point is that making the matching facility too complex may be counter 
productive. I'm in favour of very simple systems which behave as expected. 


	Phill

PS Please don't send me your perl scripts.
Received on Friday, 23 February 1996 16:39:04 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:46 EDT