Kimbro Staken wrote:

> The reason I'm asking is because there has been very little activity
> based on the last draft that was posted. I expected that to at least
> spark some comment or maybe even cries of outrage but it has been very
> quiet instead.

Sorry about that -- nothing is worse than complete silence.

I'm usually so busy that I only have time for localized bursts of
activity on XML:DB. I've only had time to glance at the latest draft,
but it looked reasonable to me. I'll try to look closely later this
week.

> I'll state the current goal of the API is to provide the ability to
> build applications that can move easily from one vendors product to
> another.

Exactly.

> Now times have changed a lot since both ODBC and JDBC were developed and
> Open Source has become a very viable way of developing software to
> achieve the same goal. An example of this would be Perl DBI and SAX.

Actually, both SAX and ODBC were developed in the same way XML:DB was
developed: iterative work on the spec and implementations by a number of
different vendors, until enough experience was gained to label the spec
1.0.

> So
> what I wanted to throw out was the question of would it be of more value
> to us to focus more on implementation and less on specification?

I think so. If we get several different implementations over several
different databases, it should ferret out any problems.

> What I
> mean is should the output of this project be not only a driver manager
> and specification but also the drivers? One stated goal of XML:DB is to
> develop specs and reference implementations of those specs so this isn't
> inconsistent.

I don't think the project itself needs to be a delivery vehicle for
drivers. I think it just needs to deliver the spec and any common code,
such as the driver managers in JDBC or ODBC. However, I do think that
drivers need to be written. A spec without implementations to prove it
is not useful.

> We're so early in the XML database game and everyone's implementations
> are so immature and so different that a comprehensive spec is going to
> be very difficult to develop. Basically I'm thinking a simple usable API
> with a few drivers in 3 months is much more valuable then waiting a year
> for us to fully detail a spec and then get some initial implementations.

Agreed.

One useful measure of success is getting a handful (three to five) of
independently-written applications to work with a handful
independently-written of drivers. If we can do this, then I'd say we
have a pretty useable spec.

-- Ron

----------------------------------------------------------------------
Post a message:         mailto:[EMAIL PROTECTED]
Unsubscribe:            mailto:[EMAIL PROTECTED]
Contact adminstrator:   mailto:[EMAIL PROTECTED]
Read archived messages: http://archive.xmldb.org/
----------------------------------------------------------------------

Reply via email to