Next message: jamsden@us.ibm.com: "Re: "stable" href's"
Date: Fri, 21 Jan 2000 23:42:08 -0500
Message-Id: <10001220442.AA26934@tantalum>
From: "Geoffrey M. Clemm" <geoffrey.clemm@rational.com>
To: ietf-dav-versioning@w3.org
Subject: Re: "stable" href's
From: "Eric Sedlar" <esedlar@us.oracle.com>
I've been thinking about the idea of a "fixed" binding within a
collection (a boolean property associated with a link) for caching
purposes. Does it really have to be an entire URL?
From: Geoffrey Clemm
If only the binding name is "imMOVEable", and not the whole URL, does
that provide the client with much benefit? It can cache ../foo
relative names between members of that collection, but it wouldn't
provide stable references for resources outside of that collection.
From: "Eric Sedlar" <esedlar@us.oracle.com>
Sure it does. Take the example of a compound document (made of multiple
resources), like a book, modelled as a collection and the set of components
in it.. I might want to ensure that the "chapter.html" never gets renamed
within the book collection. I can move the book around, but once I've found
the book, I can get to a particular component with a well-known pathname.
You can always get fix a set of bindings to fix an entire URL.
If you know what collection it is in, it's easy/efficient to find something
by using PROPFIND to look for something by its DAV:resourceid, so it is
not very critical to freeze the binding name. It is when you don't know where
the collection is that you need a stable name, and having just the last
segment be stable doesn't help.
BTW, what did you have in mind when you said "associate a boolean property
with the link"? Since a binding is not a resource, you can't put a property
on the binding, but you could put some property on the collection.
Cheers,
Geoff