XOM API for Processing XML with Java

[Tatu Saloranta <cowtowncoder AT yahoo.com>]

--- On Wed, 12/31/08, Peter Murray-Rust <pm286 AT cam.ac.uk> wrote:

...
> Tatu Saloranta wrote:
> > Just to be clear, based on error message the problem
> is not the scheme part, but that DTD references can not be located.
> 
> I guessed this might be the case. It's obviously a confusing error 
> message but that may be inevitable if Xerces is involved.

Agreed.

> Just to  confirm, the DTD itself is accessible - it is presumably
> references  within it that give the problem.

Ok, yes. That is even more confusing.

One thing you could try too, btw, maybe just to troubleshoot the problem 
itself, would be to use Nux extensions that allow building XOM document using 
Woodstox stax parser (I don't think XOM core [yet?] supports stax as document 
build source -- might be wrong here).
This because Woodstox tries to give more meaningful error messages for DTD  
handling problems.

...
> Understood. However configuring Xerces is not always a relaxing 
> operation and it would be nice to avoid it.

Yes, this does sound like an area where it would be nice to offer some 
convenience access.

... it is also nice that this is the place where I can pass the buck since 
Elliotte is the man wrt XOM. :-)
Alternatively, Nux could perhaps also offer some of such convenience settings 
(it may already do -- it has been a while since I used it).

...
> Again I guessed this - I tried standalone="yes" without success and 
> assumed some default overrid it.

This is something that I never fully understood, I mean why Xerces does not 
make use of this information. In fact, I found that some xml test suites 
count on such behavior to be the norm (I think that was the "xml test" test 
suite), whether that was intended or not.

> > 
> > Hope this helps,
> > 
> It does.
> 
> A gentle suggestion would be to be able to configure XOM to hide some of 
> this. Basically "I know what I want to do and it uses simple XML ideas 
> and doesn't need the complexity of DTDs, Schemas or whatever. I'll take 
> my chances on in working as I would like". For many of us (as I have 
> argued consistently since XML started) Xerces and XMLSchema are vastly 
> more complex than we require and they cause us more problems than they 
> solve. If it weren't for XOM I would have great
> problems in using XML.

I agree. Especially since for most users distinction between tree model (like 
XOM) and underlying parser core is irrelevant,

-+ Tatu +-