Michel Fortin wrote:
I've started a wiki page about the common subset:
<http://wiki.whatwg.org/wiki/Common_Subset>
I'd like to explore this from a different angle.
Libraries (like html5lib) will likely provide a means to serialize a
DOM, and will presumably have unit tests.
The question is: does it make sense to standardize what such a method
produces? HTML allows variations on the case of elements, single vs.
double vs. no quoting of attributes, etc.
If such were standardized, how would the HTML5 canonical serialization
differ from the XHTML5 canonical serialization (in fact, must they be
different at all?)
In any case, a desirable feature of such a serialization would be the
ability to round trip. For HTML5, this would only apply to all valid
HTML5 documents: as an example, one could artificially produce a DOM
which contains a <h1> inside the <head> element; if such a DOM were
serialized and then parsed by an HTML5 parser, the DOM produced would
differ, as well it should.
If there is no interest in standardizing a serialization (or separate
standard serializations form HTML5 and XHTML5), then this discussion
belongs on [EMAIL PROTECTED] mailing list.
- Sam Ruby
