W3C home > Mailing lists > Public > www-tag@w3.org > October 2003

Re: Action item on syntax-based interoperability

From: Tim Bray <tbray@textuality.com>
Date: Fri, 24 Oct 2003 16:10:46 -0700
Cc: "Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>, <www-tag@w3.org>
To: "Joshua Allen" <joshuaa@microsoft.com>
Message-Id: <4C8DC7C8-0677-11D8-A7A4-000A95A51C9E@textuality.com>

On Friday, October 24, 2003, at 12:44  PM, Joshua Allen wrote:

>> Whoa!  Are you getting confused about what people mean by "data 
>> model"?
> RPC/DCOM interop was all about shared marshalling formats, and shared
> interface definitions -- not about shared data models.  Nobody in the
> DCOM/Corba/RPC worlds deals with things at the data model level.

Granted; although I tend to see APIs and data models as two sides of 
the same coin, especially when thinking in O-O terms.  For example, the 
DOM is actually an API but nobody feels uncomfortable calling it an 
object model.

> As an example of an "interoperable data model", think of the relational
> data model.  The existence of the relational data model means that I 
> can
> learn SQL and use any of a number of great database engines.  The
> existence of relational data model means that I can convert my code 
> from
> using SQL Server to use MySQL with no code changes necessary --
> regardless of which programming language or API (ODBC, JDBC, OLEDB) I
> chose to use.

Really?  I have much personal bitter experience with porting apps from 
one database vendor's SQL to another's, and have observed serious 
breakage to be the rule rather than the exception.

> The benefits of a shared data model in the case of relational are
> completely undisputable.  And the benefits of a shared semistructured
> data model are already evident in enabling deployment of XPath, and 
> will
> soon be far more evident in deployment of XQuery.

I agree entirely.  But what does this have to do specially with 
*networked* information systems in general and the Web in particular?  
The virtues of SQL are perfectly apparent when you're doing a master 
file update in COBOL under whatever IBM calls MVS these days.  I assume 
the same will be true of XQuery.  The Webarch document needs to 
highlight those architectural virtues that are particularly important 
in constructing *networked* information systems, and prominent among 
those is winning interoperability by defining the bits on the wire.

Remember, this is about the architecture of the Web, not a general 
collection of sermons about goodness in technologies that are modern, 
or that W3C members use, etc. -Tim
Received on Friday, 24 October 2003 19:17:30 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:32:40 UTC