XOM API for Processing XML with Java

[Elliotte Rusty Harold <elharo AT metalab.unc.edu>]

Wolfgang Hoschek wrote:
> xom-1.1-CVS with Crimson JDK 1.4 (but not xerces-2.6.2) has a bug when 
> using doc.setInternalDTDSubset(doc. getInternalDTDSubset()) with the 
> document below. Interestingly, the document parses fine when using 
> crimson with new Builder().build(...). I'm wondering if anything could 
> be done about it?

I can't reproduce this one with the following test:

    public void testSetInternalDTDSubsetWithCrimson() {

        DocType doctype = new DocType("root");
        String dtd = " <!ELEMENT doc (#PCDATA|a|b|c)*>\n"
        + "  <!ATTLIST doc id ID #REQUIRED>\n"
        + "  <!ELEMENT a (#PCDATA|a|b|c)*>\n"
        + "  <!ATTLIST a id ID #REQUIRED>\n"
        + "  <!ELEMENT b (#PCDATA|a|b|c)*>\n"
        + "  <!ATTLIST b id ID #REQUIRED>\n"
        + "  <!ELEMENT c (#PCDATA|a|b|c)*>\n"
        + "  <!ATTLIST c id ID #REQUIRED >\n";
        doctype.setInternalDTDSubset(dtd);
        assertEquals(dtd, doctype.getInternalDTDSubset());

    }

Given that this is likely a Crimson bug, and that if it is I don't see 
any obvious way to work around it, I'm inclined not to worry about it 
too much. If you could figure out a way to write a unit test that 
exposes it (especially if it doesn't require me to manually hack the 
Verifier class just to load Crimson there); I could take another look. 
However, I suspect I should just blame this on Crimson weirdness, and 
tell people to use Xerces if they encounter it.

--
Elliotte Rusty Harold  elharo AT metalab.unc.edu
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim