Gunther Schadow wrote:
> Having come back to searching for good stuff on the net in intervals
> of about 6 months for the last 3 years I would now ask you to help
> me out here: I can't find these open source tools. Unless you have
there are quite a few. Have a look at comp.databases.object to find out
about
them. Also the Cetus object links server is very useful for findiing
things.
For our GEHR effort, we are using the matisse database which is not
free, but is
cheap, and is a rock-solid ODB (I have just finished a commercial
investment
system at a major insurer in Australia, using Eiffel/Matisse in a
production
environment). It has Java, C++ bindings. You can also look at O2 and
Versant, for
reliable databases. There are others as well, but I have no experience.
> provided me with a link to something really useful, I'm going
> to maintain that there is no such a thing and we have to craft
> something ourselves.
I admire your willingness, but I can tell you that building a decent ODB
is
definitely not for the faint-hearted, especially when you can go and buy
one such
as Matisse for USD 500(ish).
> I came to believe that the most workable option is (still) to use
> an SQL data base like PostgreSQL and hand-craft an object-relational
> wrapper around it. This is an issue we have to address for this
You will find this is an expensive operation - the mapping is difficult
to
define, usually, you often end up compromising your object model, and
the
software result can be (very) inefficient. If you really want to (have
to) use an
RDBMS as an EHR server, the thing you want to get right is the design of
the
schema. There is an approach to doing this that makes it
object-friendly, and
will allow you to write a less complicated mapping layer.
> Open EMR kernel (sorry, it is a good term, see Webster's) So please,
> David, in all your enthusiasm for good system design, point us to
> something real, code, stuff that you can really use now.
Just to reiterate, the Ocean GEHR kernel will be ready by about
mid-2000, and
early versions will be available as soon as we get the source part of
the site
set up on www.gehr.org.
We will be using UML to make people happy, by the way, so that it is
less
difficult to discuss the diagrams.
regards,
- thomas beale
