Quoting Sunil Mishra (smishra@sfmishras.com):
CL-USER(116): (dom:map-document (cxml:make-namespace-normalizer (cxml:make-octet-stream-sink *standard-output*)) *)
Note that make-octet-stream-sink defaults to canonical mode for historical reasons.
<svg xmlns="http://www.w3.org/2000/svg"> <script type="text/css"> </script> </svg> #<MULTIVALENT stream socket connected from localhost/3813 to localhost/3817 @ #x205003d2>
Sorry, I don't see a bug. The serializer in canonical mode outputs character references for the newlines here, but it doesn't output a CDATA section either in the first place, so that's fine.
If you want to see a CDATA section, use non-canonical mode:
cl-user(43): (dom:map-document (cxml:make-octet-stream-sink *standard-output* :canonical nil) (cxml:parse-file "~/graph.xml" (cxml-dom:make-dom-builder))) <?xml version="1.0" encoding="UTF-8"?> <svg> <script type="text/css"> <![CDATA[
]]> </script> </svg>
``Within a CDATA section, only the CDEnd string is recognized as markup, so that left angle brackets and ampersands may occur in their literal form; they need not (and cannot) be escaped using "<" and "&". CDATA sections cannot nest.''
Can cxml please correctly follow this requirement?
It follows this requirement while parsing.
Only in serialization there is one little "problem" (unrelated to your question):
A document constructed in memory might include a CDATA section with characters not representable in a CDATA section. That is a user error, and CXML should signal an error when told to serialize such a document in non-canonical mode; right now I believe it does not signal that error and outputs the user data as-is, resulting in output that isn't well-formed. (But I'm taking patches. :-))
d.