Re: XML and required DTDs

Paul Grosso wrote:

> The scenario I see is as follows:  my customer uses Adept to create an
> XML compliant document that conforms to a DTD and they send that to
> someone else who uses a public domain XML editor that doesn't use
> DTDs.  That other person will modify the document in the XML editor and
> will very likely do something so that the document no longer complies
> with the DTD my customer used to create it.  Then the other person
> returns the document to my customer who tries to bring it up in Adept
> and can't.  Or, maybe Adept is smart enough in "XML mode" to
> auto-provide declarations for undeclared elements and to ignore
> contexts that are not permitted by the DTD.  Then my customer goes to
> re-insert that document into their SGML repository, the the database
> integrated with Adept can't accept the invalid document.
> 
> Does all this give either my products and/or SGML itself a black eye?

Paul, what is the difference between your first scenario and one in
*real SGML*
(gad, i disklike that phrase), in which an instance arrives with or
without 
a DTD, the author opens it in a text editor, makes non-conforming 
changes, then returns it without parsing it?  IOW, isn't that a 
process glitch and is actually the common case?  It is why we teach 
them to ensure two transmission sites are using the same version of 
a DTD.  SGML gets a black eye in this case only if the users aren't
really analysing the problem.

The case is different if XML explicitly allows dynamic instancing of new 
declarations in the presense of a DTD sent by the original author.  In 
that case, the user MUST return the modified DTD and notify the receiver 
that it has been modified if that modification requires adjustment of 
any device (database, filter, etc) that processes the information.
XML gets a black eye if the XML vendor doesn't tell the customer what
the consequences of using the dynamic features can be.  In that case, 
yes, it is a marketing/training problem.

Am I understanding you correctly?

len bullard

Received on Saturday, 14 September 1996 14:54:14 UTC