W3C home > Mailing lists > Public > public-webarch-comments@w3.org > July to September 2004

Re: when to use XML, RFC3117 [was: Review of Web Architecture doc - WD-webarch-20031209]

From: Roy T.Fielding <fielding@gbiv.com>
Date: Fri, 30 Jul 2004 16:46:31 -0700
Cc: public-webarch-comments@w3.org
To: Dan Connolly <connolly@w3.org>
Message-Id: <AF236388-E282-11D8-BF2D-000393753936@gbiv.com>

I have no objection to including it as a reference, but for the
record: I don't find the text to be useful or compelling, and
further believe that BXXP suffered from the choice of XML as a
default packaging syntax in the same ways that SOAP has suffered
from that choice.  That is why that choice was overturned during
the standardization of BEEP (see section 6).

Marshall's original design choice was based on somebody else's
deployment model (i.e., the availability of non-secure XML
parsers for authoring), rather than a deployment model specific
to the needs of the protocol.  That is not surprising, given that
he designed BXXP as an exercise in generic application framing
protocols rather than as an application in need of deployment.

The discussion of SNMP's use of ASN.1 is particularly telling,
since that was one of the most aggravating aspects of SNMP and
the subject of many debates.  Again, the only reason it was
axiomatic is because a choice was made by the original designer
and there was no compelling need to switch encodings once that
choice was made, even though there was ample evidence that
ASN.1 made deployment of SNMP very difficult.

In general, I don't respect the use of "axiomatic" in a design
description.  What it means is that the designer has no compelling
argument (or principle) upon which to base the design choice,
and is merely calling it axiomatic because they have no desire to
argue the point.

....Roy
Received on Friday, 30 July 2004 19:45:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:26:47 UTC