Re: When parsing speed matters (was Re: [xml-dev] No XML Binaries? Buy H

[Boris Kolpackov <boris@codesynthesis.com>]

Hi,

Richard Salz <rsalz@us.ibm.com> writes:

> It can be a pretty compelling argument  when you tell someone that
> they can do full XML security, schema validation, threat prevention,
> etc., with absolutely no additional cost to their application
> deployment.

I think a more compelling argument would be the speedup of the
application itself. An appliance does XML parsing and schema
validation (which includes a major chunk of data extraction
[what Noah calls deserialization] and dispatching) but then
it throws all that expensive work away when instead it could
pack it all up in an efficient, schema-aware binary format
and forward that to the application.

I did a quick test to verify this. Our data binding product can
save the object model in CDR (binary data format used in CORBA).
I measured the time it takes to extract that object model from such
a CDR representation against parsing the original XML using Expat
with do-nothing SAX handlers. This is not really a fair comparison
since the CDR test creates a complete object model which an
application can use right away while the SAX test lacks the data
extraction stage.

The XML instance (300Kb) and XML Schema were taken from a real
application developed by one of our clients. It contains mostly
strings which also favors the SAX test. The comparison shows
that the CDR extraction and object model instantiation is almost
twice as fast as just XML parsing.


-boris


-- 
Boris Kolpackov
Code Synthesis Tools CC
http://www.codesynthesis.com
Open-Source, Cross-Platform C++ XML Data Binding