- From: Yaron Goland <yarong@microsoft.com>
- Date: Thu, 31 Oct 1996 15:02:40 -0800
- To: "'Larry Masinter'" <masinter@parc.xerox.com>, "'connolly@beach.w3.org'" <connolly@beach.w3.org>
- Cc: "'ejw@rome.ics.uci.edu'" <ejw@rome.ics.uci.edu>, "'w3c-dist-auth@w3.org'" <w3c-dist-auth@w3.org>
I realize that my new ranges retrieval idea isn't very well baked but let me give my motivation. I want to retrieve the AttributeDirectory for a resource but the directory is HUGE. My poor little WebTV box can't handle that much memory. I need a way to just pull in a part of the document. I can't use byte-ranges because I might end up retrieving half of an entry. Imagine an entry in the AttributeDirectory is "sitemap = http:\foo\bar\blah". However my byte-range request cut the entry so I only received "sitemap = http:\foo\bar". The only way I can know there is a problem is if I request another chunk of the file. This problem also applies to using DAV for newsgroups or e-mail. The truth is that we are running into a database problem. How do we retrieve specific records? I have a feeling that this is a matter for another group. I know that there has been a lot of talk about having standardized database access over HTTP around here at MS. Has there already been a W3C or IETF workgroup on the subject? Yaron >-----Original Message----- >From: Larry Masinter [SMTP:masinter@parc.xerox.com] >Sent: Thursday, October 31, 1996 10:53 AM >To: connolly@beach.w3.org >Cc: Yaron Goland; ejw@rome.ics.uci.edu; w3c-dist-auth@w3.org >Subject: Re: Prelim. DAV spec. > >> A representation (or entity -- same thing) is immutable, the way >> integers and URLs are immutable. How many versions of the number >> 2 or the string "http://www.w3.org/" are there? > >Well, a 'representation' isn't exactly the 'same thing' as an entity. >We're working in an area where there aren't enough precise terms. > >It's very risky to try to make a 'plain logic' argument where we're >primarily having difficulty with definitions of terms. > >Let's try this out along a different dimension than content >negotiation: > > >HTTP supports 'range retrieval' as well as content negotiation. >Let's suppose that we allow range retrieval for 'pages' out of a >single multi-page resource: > > request: > Range: pages=1,2 > >without giving separate URLs for each page independently. There's one >URL: > http://my.server.dom/book/chapter1.pdf > >but I can either retrieve the whole thing or several pages at a time. >One might imagine wanting to 'version' the pages independently, e.g., >"version 12 of page 7" and "version 9 of page 12" and even have some >way of saying "version 10 of the chapter consists of version 23 of >page 1, ... version 12 of page 7 ... version 9 of page 12 ....". > >This isn't logically inconsistent, it's just implementationally >difficult. > >At least in the context of versioning, the part/whole relationship >bears a strong resemblance to the representation/resource >relationship. > >Larry > > > > > >
Received on Thursday, 31 October 1996 18:02:42 UTC