* Matt Sergeant ([EMAIL PROTECTED]) wrote:
> On Wed, 14 Nov 2001, Stephen Adkins wrote:
> >
> > * Is anyone interested in helping sort out Date and Time
> > functionality for the P5EE?
>
> No. This is *not* an enterprise API, IMO. Others may feel differently, but
> we're not here to sort out all of CPAN!
>
I agree with Matt, we need to keep this project's scope confined. So
if I may (i'm going to do it anyway) I'd like to say what we should be
looking at.
Enterprise computing has a constantly changing definition, however
currently enterprise computing is generally reguarded as 3 tier
computing[1]. The middle tier being for business logic. The other two are
for UI and system services. And it is this middle tier we have to keep
tightly focussed on - i would even argue that templating is outside of
the scope of this project, thats UI stuff. As for the discussions on
pod extensions, for me the jury is still out on that one.
The sort of things we should be looking at are ...
Persistent transactions
Session storage
Message based systems
Persistent object ``brokers''
Remote method calling
Services and service location
especially with reguards to how these technologies can be useful for
describing business logic (remember its not enough to present the
technologies in a standard fashion from a bottom up view, they need to
be appealing to application architects as well). Now sometimes, such
as with error handling we will have to go a little deeper and
standardise, but we definetly don't need to do it with date and
time. I'd also suggest that once people start writing the code then we
will know what requirements we have for areas such as error handling
and until then it is all guess work.
I'm not saying that looking at date and time is not worthwhile, I'm
simply saying we need to stay focused. If people really want to create
a date and time API, then great go and do it, but don't do it here -
this project already has a high chance of failure without loading it
with more work.
What I believe we are experiencing is a side effect from the delays in
setting up the CVS server. While we don't have a CVS server we tend to
discuss things ad infinitum and it is only natural to come up with new
topics for discussion. Once we get the CVS server operational it will
either be time to shit or get off the pot. If you take my meaning ;-)
Greg
[1] I believe that things like .Net only distribute these tiers wider
and involve distribution beyond the limits of a single organisation
(i.e. MS providing passport, is effectively a system service with a
nice business logic friendly layer on top of it).
--
Greg McCarroll http://217.34.97.146/~gem/
