Okay, let's start from the beginning and build from there 
as we discuss what features we need in a datatype validator 
implementation. First, some requirements. Granted, these
are *my* requirements but I will use them in the course of
my design discussions.

REQ #1: Validate DTD and XML Schema datatypes.
REQ #2: Share datatype validators among multiple grammars.
REQ #3: Be able to interchange datatype library.
REQ #4: Provide standalone implementation.
REQ #5: Allow applications to define new datatypes.

Requirement #1 is simple enough because we know what 
datatypes that DTDs and XML Schemas provide. The challenge
is to design a datatype validation framework that is not
directly tied to either one so that we can we don't
introduce unneeded dependencies (see REQ #4) and is
extensible (related to but going beyond REQ #5).

Requirement #2 is very important. In the design for the
entire validation engine, we have to think about being
able to support grammar caching. This means that every-
thing should be able to shared among parsers in different 
threads. So whatever we design needs to take this into
account.

Requirement #3-5 are more like my wish list. These are
not absolutely required but I think that we can support
some really cool features if we focus our design on 
allowing this functionality. And I don't see anything
that would prevent us from accomplishing this.

I have a habit of being long-winded in posts so I'll
post a series of small messages that address different
aspects of the problem and make design proposals
accordingly.

-- 
Andy Clark * IBM, TRL - Japan * [EMAIL PROTECTED]

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to