[office-comment] RELAX NG tools and OfficeOpen.Org DTD

[John Cowan <jcowan@reutershealth.com>]

A few points:

1) I processed the Sun submission through James Clark's conversion program
DTDinst (http://www.thaiopensource.com/dtdinst), which converted it mostly
successfully.  I flushed one bug in DTDinst, which I have reported: it
correctly creates an office.rng file corresponding to the office.dtd
file, but it overwrites the *.mod files with RELAX NG equivalents.
This bug is trivial to work around, though.

Two XML Namespaces prefixes were reported as undeclared: "presentation" in
drawing.mod and style.mod, and "dialog" in dialog.dtd.  After declaring
these, the Jing validator reported no errors in the converted RELAX
NG output.

2) The current state of James's conversion tools, all of which attempt
to preserve the structure of schemas, is as follows:

	conversion from RNG (XML syntax) to RNG (compact syntax): exact
	conversion from RNG (compact syntax) to RNG (XML syntax): exact
	conversion from DTD to RNG (XML syntax):
		does not convert general entities or notations (no equivalent)
		converts parameter entities to RNG definitions
	conversion from RNG (either syntax) to DTD:
		does not handle all RNG constructs
		often generates DTDs that are looser than the RNG
		roundtrips DTD > RNG > DTD conversions correctly
	conversion from RNG (either syntax) to WXS:
		does not handle all RNG constructs
		often generates WXSes that are looser than the RNG
		output may violate certain WXS rules
	conversion from WXS to RNG: not available

3) I recommend that the TC *not* loosen "arbitrary" sequences to RNG
"interleave" or WXS "all".  That provides unneeded syntactic flexibility
and complicates programs that must interpret the syntax, while not
helping programs that generate it.

-- 
What is the sound of Perl?  Is it not the       John Cowan
sound of a [Ww]all that people have stopped     jcowan@reutershealth.com
banging their head against?  --Larry            http://www.ccil.org/~cowan