[
Lists Home |
Date Index |
Thread Index
]
On Thu, 2004-12-16 at 17:52 +0000, Bill de hÓra wrote:
> Only half-jokingly, this is what applications are for! I wouldn't
> generally build staleness like the above into data as it narrows
> potential usage, never mind questions like "90 days since when?".
>
> What I have seen is that the data (ie an XML message) or the event (ie a
> birth), or both, are timestamped and endpoints usually have enough
> information to go on. Even then it gets complicated - I've seen
> incidents arise due to the time lag between the time of an event and the
> time the representation of that event hit the wire.
As Mike Kay says then, it is a very real problem.
Dealing with 1553B messages would probably disallow XML anyway,
but for more general usage, it would seem that a snapshot property
would add something to Rogers model? Knowing the dateTime of the
snapshot would.. might provide enough for the Pareto?
It would certainly account for items such as employee (today at least),
haircolor... if aged over 21 (maybe :-)
Certainly worth considering when modelling. Even if the result is
to say as Bill indicates, that the model just isn't timely enough?
--
Regards,
Dave Pawson
XSLT + Docbook FAQ
http://www.dpawson.co.uk
|
