[xsd-users] Fast InfoSet or other binary format?

[Boris Kolpackov]

Hi Ray,

Ray Lischner <rlischner at proteus-technologies.com> writes:

> How easy would it be to support Fast Infoset? How about a different
> binary format that requires knowledge of the schema structure? (I
> already know about the low-level custom streaming. It's nice, but
> I need to investigate formats that preserve the structure. Fast
> Infoset is one possibility, but I am considering other formats, too.)

By structure I assume you mean some sort of type information in the
broadest sense of this word (e.g., XML tags are type information). We
can then divide all formats into two categories: the ones that are
equivalent to or contain less type information than XML. Fast Infoset
is one such format. The second category of formats require type
information that is not available in XML so one would need access
to XML Schema.

I think the first category of formats is best handled at XML parsing
level by either emulating SAX events or building DOM documents directly
(or both: SAX for parsing and DOM for serialization). The nice side
of this approach is that it is not vocabulary-specific so the same
code can used for different things. I think it should be pretty
straightforward sans the complexity of the format.

For the second category of formats I don't see any other way than
writing your own code generator. This should also be quite straight-
forward from the code generation side.

We can help with either type of work.


hth,
-boris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 652 bytes
Desc: Digital signature
Url : http://codesynthesis.com/pipermail/xsd-users/attachments/20070214/a51655db/attachment.pgp