Imported from previous forum
[ original email was from sam johnson - sam.johnson@gs.com ]
I’ve been doing some experimenting with XML and SGML both, and it seems that a major limitation of deriving “FIXML” from XML is that XML limits what can be done with datatypes.
Someone correct me if I’m wrong here: One of the major advantages of being able to specify the message definitions in a DTD is that both format and “datatypes” are well-defined and validated by the parser; however, XML seems to restrict pretty severely what can be done with primitive data.
Of course, it’s entirely possible I’m missing something. For example, I’d like to be able to define an ELEMENT FIXDate which is of format YYYYMMDD (all numeric) and validated by the DTD. This is critical, I think, and pretty straightforward in SGML but I can’t seem to do it in XML. Anyone else played with this?
[ original email was from John Goeller - jgoeller@sbi.com ]
> I’ve been doing some experimenting with XML and SGML both, and it seems that a major limitation of deriving “FIXML” from XML is that XML limits what can be done with datatypes.
>
> Someone correct me if I’m wrong here: One of the major advantages of being able to specify the message definitions in a DTD is that both format and “datatypes” are well-defined and validated by the parser; however, XML seems to restrict pretty severely what can be done with primitive data.
>
> Of course, it’s entirely possible I’m missing something. For example, I’d like to be able to define an ELEMENT FIXDate which is of format YYYYMMDD (all numeric) and validated by the DTD. This is critical, I think, and pretty straightforward in SGML but I can’t seem to do it in XML. Anyone else played with this?
>
>
Have you read the XML-Data proposal?
www.w3.org/TR/1998/NOTE-XML-data-0105/Overview.html
It is introducing the concept of schemas which will provide the same functionality as DTDs. Schemas are written in XML so they are extensible. Developers will be able to extend schemas with additional information like datatypes.
See also:
www.textuality.com/xml/typing.html
[ original email was from John O’Hara - john.o’hara@bankerstrust.com ]
>
> Have you read the XML-Data proposal?
> www.w3.org/TR/1998/NOTE-XML-data-0105/Overview.html
>
> It is introducing the concept of schemas which will provide the same functionality as DTDs. Schemas are written in XML so they are extensible. Developers will be able to extend schemas with additional information like datatypes.
>
> See also:
> www.textuality.com/xml/typing.html
>
There are several competing proposals for how to implement some kind of document ‘schema’ for XML to overcome the weaknesses of DTDs.
The XML-Data proposal was one of the first, but unlikely to be accepted as the final standard (that’s the general feeling I get from the XML community) DCD or DDML is more likely to be adopted.
However, lack of complete validation through the DTD is no reason to rule out XML processing for FIX; validation can be done in the processing API/engine until something better comes along.
What FIXML does open up, as I’m sure everyone involved appreciates, is the ability to leverage existing and emergent tools and a well understood idiom for the messages. There are other benefits; easily understood by people and perhaps more importantly, easily extended with and embedded into other data as future requirements grow.
Finally XML notations are flexible enough to represent complex recursive message structures; which would be required to represent some complex exotic products if FIX continues to move to support the widest range of financial products…
Regards,
John O’Hara
Technical Architect
Bankers Trust
[ original email was from sam johnson - sam.johnson@gs.com ]
this is a test reply…
wish me luck.