Quoting Peter K.Lee ([EMAIL PROTECTED]):
> I must have grabbed the "UTF-8 w/o Unicode" from the Recent Changes
> section, erroneously condensing the part about "Lisp Implementations
> w/o Unicode support" into a garbage phrase. :)
Well, that changelog entry was about yet another issue. What *internal*
representation of characters and strings does the parser use? On Lisps
with Unicode support is the answer is obvious. But what does a program
do that wants to process Unicode when users will want to run it on Lisps
that don't support Unicode?
Closure's answer was a radical one: Since the free Lisps did not have
Unicode support when it was written, Gilbert simply used character codes
and (unsigned-byte 16) arrays instead of characters and strings, and
calls them runes and rods.
CXML still the concept of runes and rodes, but on Lisps with Unicode
support they are real characters and strings. On Lisps without Unicode
support, above-mentioned rods are still in use internally, but after
getting "bug" reports about CXML using code points instead of characters
on CMUCL, I gave in and made recoding to UTF-8 the default.
So the internal representation of strings on non-Unicode-aware Lisps is
that of Lisp strings containing characters that are really UTF-8
bytes.
I don't think it would be terribly interesting to explain all this in a
superficial comparision of XML parser implementations, but it is an
interesting implementation problem for Lisp programs that aspire to
portability in general.
> Soon I hope to have actual conformance testing taken place on each XML
> parser library and have the results reflected in the comparison
> report.
>
> In the meantime, I hope the omission of conformance is forgivable. I
> do value the importance of that factor in properly evaluating the
> completeness of the parsers.
No need to run them yourself in most cases. Reproducing whatever
conformance claim the parser's documentation makes could be enough.
CL-XML documentation:
The processor passes 1749 of the 1812 tests
(more or less all)
CXML's XMLCONF file:
0/1829 tests failed; 333 tests were skipped
(more or less all)
s-xml:
States explicitly that XML is not fully supported. ("This XML parser
implementation has the following limitations:")
XMLS:
"Understands enough of the xml spec to parse many common documents"
"Parses 85 out of the 98 valid documents in the oasis parser compliance
suite."
(so what about the other 1727 tests?)
XMLisp:
that's the only one without a statement on conformance
d.
_______________________________________________
Gardeners mailing list
[email protected]
http://www.lispniks.com/mailman/listinfo/gardeners
