[jdom-interest] Toward beta 9

[Alex Rosen]

> My sense is that this has been resolved long ago. Correctness is not 
> negotiable. JDOM will verify.

I don't know where you got that impression. Go back and read the thead
I just referenced, in particular Jason's message at the end. Unless
there was another discussion afterwards, it was still very much up in
the air, with me on one side, you on the other, and Jason in the
middle.


> It's not a question of protecting programmers from theirselves. It's

> a question of whether JDOM is correct or incorrect. If JDOM is an XML

> object model, it has to model XML, not some superset thereof. That 
> means verifying XML's constraints. 

This is simply not true. If JDOM were a parser then that would be true,
but JDOM is not a parser. Section 1 of the XML spec makes it clear that
the requirements that it lays out are for parsers only. When you're
creating and writing XML, you can choose whether to be strict or
lenient, and for good reason - different people will have different
requirements.

It's simply a fact of life that efficiency and sanity-checking must be
traded off against each other. You seem to be proposing that
sanity-checking trumps efficiency - we need to verify *everything*, no
matter what the cost. How come?

Alex Rosen
Novell, Inc.