Re: XML protocol comparisons

Eric - this is very handy! Couple of ideas for more facets/fields,
protocols, scope...
 
discussion link:
 a link to discussion archives, mailing list etc where this exists
(eg. I've seen messages cc:'d between here and a SOAP list somewhere --
I don't know if it is archived and available though). 

Graph encoding?:
I know at least the SOAP and BizTalk stuff appears to use
XML as a transport for directed labeled graphs (I have in mind the paper
at http://www.biztalk.org/Resources/canonical.asp rather than the one you
link to). If some of the others are using XML in this fashion it'd be
interesting to track this.


Other Protocols / Scope:
It's interesting to know how to scope this. XML for protocols is clearly
the focus, but then there are some outputs from the HTTP-NG investigations 
 (http://www.w3.org/Protocols/HTTP-NG/) which I'd consider relevant. They
proposed a layered approach which I believe used a binary encoding by
default but allowed an XML-based format to be dropped in. Seems a shame to
omit (as 'not an XML protocol') approaches that adopt such a layered
approach, but then HTTP-NG _wasn't_ an XML-based protocol proposal. Maybe
a separate listing somewhere to at least bookmark these?


Similarly for the tuplespace-based systems (Linda, Javaspaces,
T-Space etc. [1]). They've something to contribute to the debate but
aren't explicitly XML-based (although the various Java-2-XML mappings may blur
this distinction somewhat). For example I try to scan the Javaspaces list
periodically and noticed [2] that T-Spaces apparently has XML support:

	"TSpaces is a database system with a tuplespace front end.   It has
	persistent data, queries, multi-read and
	multi-write operators, XML support and event notification based on
	database queries, template matching or XML queries (an XQL subset)"
  
A reply from someone at Sun about Javaspaces [3]:

	Because we don't have complex queries we don't need to care how data is
	stored in JavaSpaces entries, so we don't have to add specific support
	for XML (or the next big thing that comes along.)  If you want to stick
	XML strings in your entries that should work fine.	

This is an interesting exchange since the two share the Linda heritage,
one calls out XML support (of some kind) while the other doesn't. Does
that qualify T-Spaces but not Javaspaces as an XML Protocol? (or do
neither count if they're using Java RMI instead of HTTP for comms?)

This is in no way a criticism of Eric's list, just thinking out loud about
the difficulty of scoping such discussions...

Dan 

[1] http://lists.w3.org/Archives/Public/xml-dist-app/2000Mar/0055.html
[2] http://archives.java.sun.com/cgi-bin/wa?A2=ind0002&L=javaspaces-users&F=&S=&P=5307
[3] http://archives.java.sun.com/cgi-bin/wa?A2=ind0002&L=javaspaces-users&F=&S=&P=6780

On Wed, 29 Mar 2000, Eric Prud'hommeaux wrote:

> I put together a comparison of a bunch of XML protocols, 
> 
> SOAP [http://www.w3.org/2000/03/29-XML-protocol-matrix#SOAP]
> ICE [http://www.w3.org/2000/03/29-XML-protocol-matrix#ICE]
> WDDX [http://www.w3.org/2000/03/29-XML-protocol-matrix#WDDX]
> BizTalk [http://www.w3.org/2000/03/29-XML-protocol-matrix#BizTalk]
> IOTP [http://www.w3.org/2000/03/29-XML-protocol-matrix#IOTP]
> TIP [http://www.w3.org/2000/03/29-XML-protocol-matrix#TIP]
> WfXML [http://www.w3.org/2000/03/29-XML-protocol-matrix#WfXML]
> ebXML [http://www.w3.org/2000/03/29-XML-protocol-matrix#ebXML]
> XMI [http://www.w3.org/2000/03/29-XML-protocol-matrix#XMI]
> 
> for everyone to discuss/dispute. It is said that the best way to get a
> question answered on usenet is to post an incorrect answer. Persuant
> to that, I have not done extensive readings of some of the protocol
> papers during my characterizations, but at least they're all there in
> a forum where we can compare apples and fruit baskets.
> 
> I'll be adding more dimensions and would like feedback on what people
> wish to compare. Also, I'd like to have anchor-rich HTML versions of
> the documents so I can point to specific parts of the spec as
> supporting evidence.
> 
> -- 
> -eric
> 
> (eric@w3.org)
>

Received on Thursday, 30 March 2000 03:51:32 UTC