W3C home > Mailing lists > Public > xml-dist-app@w3.org > September 2002

Re: [AF] relative URIs for attachments

From: Ray Whitmer <rayw@netscape.com>
Date: Tue, 17 Sep 2002 03:57:00 -0700
Message-ID: <3D870A7C.7090806@netscape.com>
To: noah_mendelsohn@us.ibm.com
CC: Henrik Frystyk Nielsen <henrikn@microsoft.com>, Herve Ruellan <ruellan@crf.canon.fr>, xml-dist-app@w3.org

noah_mendelsohn@us.ibm.com wrote:

>>>Could you elaborate on how the binding allows or
>>>disallows use of relative URIs?  How does this
>>>interoperate with XML Base?  It seems to me that if you
>>>refer to a part from within a message, xml base kicks
>>>in, and there is no problem unless there was no
>>>declaration of the base in the message.  Or does the
>>>above language permit bindings to ignore XML base?
>>>      
>>>
>
>Briefly: the binding decides how to actually package
>and/or move the attachments.  Could be
>SOAP+Attachments, could be DIME, could be something
>else.  It's up to the binding.  Since it knows where
>the attachments (formally called parts) are, it knows
>how to name them.  For example, I believe
>SOAP+Attachments [2]uses the CID: scheme, so a binding
>to S+A would reference that.
>  
>
So far, this does not seem to be answering my question about how the 
binding apparently allows or disallows use of relative URIs within 
messages.  This talks about how it may establish the URIs of the 
attachments.

>The XML Base recommendation [3] provides that:
>
>"RFC 2396 [IETF RFC 2396] provides for base URI
>information to be embedded within a document. The rules
>for determining the base URI can be summarized as
>follows (highest priority to lowest):
>
>1. The base URI is embedded in the document's content.
>
>2. The base URI is that of the encapsulating entity
>  (message, document, or none).
>
>3. The base URI is the URI used to retrieve the entity.
>
>4. The base URI is defined by the context of the application.
>
>Since the binding is what creates the encapsulating
>entity (the message with the parts), it has the
>opportunity to establish the base URI for that
>encapsulation.  So, the binding MAY establish a base;
>if it does, then relative URI's are supported.  Whether
>the binding allows base URIs to be embedded in the
>encapsulation is up to the binding; I don't believe
>  
>
>that DIME, for example, does.
>  
>
I am quite familiar with this, none of which is referenced in the 
attachment specification, nor should it be because the resolution of 
these attachment URIs should not be different from the interpretation of 
other URIs.

If the binding selects a URI syntax for attachments which permits 
relative referencing, then any application can establish the base within 
the content using XML Base and there is no reasonable way for it to stop 
the application from resolving the URIs used to refer to attachments or 
any other resources.  Attachments is a special case of general use of 
URI refererences.  It is not up to the binding to "allow" or "disallow" 
relative URI references, except in the normal part it contributes to 
their resolution, which is much less than seems presumed by the proposal.

I find fault with the specific text of the proposed modification:

"In addition, if a SOAP binding allows the use of relative URIs, it must 
specify how the base URI is established."

This goes beyond the power of the binding to allow or not allow the use 
of relative URIs.  Please show me where DIME does it.  I have no doubt 
that a chosen syntax for attachments may make relative references to 
attachments impossible, but that is quite different from what has been 
stated.  More realistic might be:

"The binding may specify a default base URI to be used in the absense of 
more-local base URIs to resolve relative URI references, including 
references to attachments."

For example, someone implementing a binding which defined no base might 
still use relative URIs by adding a base to the root element of the 
document.   Or it may choose to override the binding base with something 
else using xml base that then prevents referring to attachments by 
relative names.

Perhaps I misunderstood the proposal?

Ray Whitmer
rayw@netscape.com
Received on Tuesday, 17 September 2002 06:57:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:11 GMT