[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: Should Subject Matter Experts Determine XML Data Implement

• From: Chin Chee-Kai <cheekai@s...>
• To: "Costello, Roger L." <costello@m...>
• Date: Sat, 04 Oct 2008 21:26:37 +0800

Costello, Roger L. wrote:
> Nonetheless, my question remains: what is the connection between a Data
> Specification and a derived XML Schema design?
>   
Perhaps in a simplistic way, Data Specification is "executable" by human 
users where user can look at an incoming document and decide against the 
Data Specification whether the document is valid or not, while XML 
Schema is executable by machine in the usual manner.

But the connection is not exactly one-to-one in general.  Data 
specification tells us what the user apparently wants , but is generally 
"leaky" in that in might not have described certain exclusions or odd 
inclusions.  XML Schema says very specifically what is accepted and what 
not, but may be over-specified against the original Data Specification, 
such as the use of recursive type definitions.

XML Schema is also not the all-inclusive data filter/specifier.  Simple 
example is the Data Specification describing the input of a valid and 
correctly formated address.  Any XML Schema alone for the address type 
could not have prevented a non-existent but valid address line  (eg "123 
Mars Avenue"), which is scarcely what the user wants.  But Data 
Specification is at least "intended" to be the all-inclusive 
specification for the user.  Therein, it would seem, lies the inherent 
discrepancies when attempting to establish a connection between the two.

regards,
Chin Chee-Kai

• References:
  • Should Subject Matter Experts Determine XML Data Implementations?
    • From: "Costello, Roger L." <costello@m...>
  • Re: Should Subject Matter Experts Determine XML Data Implementations?
    • From: Dave Pawson <davep@d...>
  • RE: Should Subject Matter Experts Determine XML Data Implementations?
    • From: "Costello, Roger L." <costello@m...>