W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 1996

RE: non-HTML resources in webdav spec

From: Yaron Goland <yarong@microsoft.com>
Date: Tue, 12 Nov 1996 17:15:27 -0800
Message-ID: <c=US%a=_%p=msft%l=RED-44-MSG-961113011527Z-2576@INET-03-IMC.itg.microsoft.com>
To: "'Kenji Takahashi'" <kt@nttlabs.com>, "'w3c-dist-auth@w3.org'" <w3c-dist-auth@w3.org>
1. You need a PUT for every resource you create, HTML or not. Attributes
are resources, so every attribute you create will require a PUT. HTTP
Attributes have nothing what so ever to do with HTML.

2. It is not acceptable to have to download every single attribute every
time I do a GET. The overhead is huge. As for abusing the term header,
fine, call them attributes. Drop the word header. An attribute is
nothing more than a typed link. The semantics of this link is that when
a resource with attribute linked resources is manipulated the associated
resources are also effected. As for manipulating the name space, this is
a trade off. Performing two gets every time I want to get an attribute,
once to get the URI of the attribute and again to retrieve the
attribute, is not acceptable. Its too big a performance hit. Thus we
must have a means to refer to an attribute without two lookups. The URL
kludge is a legitimate way to do that. Though I suspect we will change
the semantics to use ? rather than the "<"...">" syntax.

3. There are no references in the document to HTML. HTML has nothing to
do with the document.


>-----Original Message-----
>From:	Kenji Takahashi [SMTP:kt@nttlabs.com]
>Sent:	Monday, November 11, 1996 9:19 PM
>To:	w3c-dist-auth@w3.org
>Cc:	kt@nttlabs.com
>Subject:	non-HTML resources in webdav spec
>Yaron (or anyone),
>I think I am coufused but...
>Do we need two PUTs to create a new non-HTML resource:
>one for attributes headers of the resource and the other for 
>the content of the resource?  
>Or the WEBDAV draft excludes non-HTML resources (although it
>does not explcitly say so)?
>To handle non-HTML resources properly, I think it would be
>better to transmit meta-information (including attributes)
>as HTTP headers.  Because you can create a resource with one
>PUT.  The draft says in 2.2.1, "attributes headers may grow to 
>very large sizes and may contain octet data", but
>I don't understand why we shoudn't send very large headers or
>if octet data are really needed?  Also sending HTTP headers in
>the body of the entities is obviously misuse by definition.
>I think clearly defined methods, such as a method for set attribute
>values, are better thatn the misuse of (or confusing usage of) 
>existing methods...  Also the draft proposes significant
>additions to URI.  I think URI should be opaque as discussed in
>the early time in this maling list (remember the "URL;ver=" 
>Actually the draft confuses me - I don't quite understand whether it 
>describes about HTTP headers or HTML head parts.
>Best regards,
Received on Tuesday, 12 November 1996 20:15:29 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:01:09 UTC