--- On Sat, 1/2/10, Uche Ogbuji <[email protected]> wrote:
On Sat, Jan 2, 2010 at 1:32 PM, MURATA Makoto (FAMILY Given)
<[email protected]> wrote:
>> It is true that there are not so many data binding tools for RELAX NG. I
>> don't think that the reason is the language design, however. JAXB V1
>> supported RELAX NG.
True, but it also supported DTD; and more important, was a very pre-mature API,
not very useful. V2 specifically does not, and at least for DTD it could not be
made to work for full round-trip (ability to generate java classes from schema,
vice versa).
...
>> In my understanding, the reason is the lack of interest (both users and
>> developers).
This certainly is an important reason.
But I think this:
> I tend to think the issue is a philosophical one. I feel RELAX NG's
> aesthetic as "text first, data incidentally." That goes well at odds
> with the idea of rigidly mapping XML structures to programmatic
> structures, which is the essence of most data binding tools.
is what I was trying to say. I do think design of RelaxNG, and its strength,
are related to challenges in using it for data binding.
XML Schema, on the other hand, has lots of weaknesses regarding "traditional
XML" (document-oriented, not data), due to its focus on data-oriented use
cases. But for data binding these are strengths.
Still: I would love to be proven wrong in this case. I added RelaxNG validation
support for Woodstox well before W3C Schema support, and love compact notation,
simplicity and power.
If there was development in this area, I would be very interested in it.
Perhaps even be able to help (I have worked a lot in data binding).
-+ Tatu +-
