Re: [Hardhats-members] == VistaWeb Missing Apps ==

[Gregory Woodhouse]

I'm just on my way out the door right now, but while I think the relational model sometimes gets a raw deal around here, I also agree that it is not the be and end all of database technology. If you try and sit down and work through the mathematics, you'll quickly find that object models have their problem (not insurmountable, IMO), too. I do think we're seeing the beginnings of a paradigm shift (and only recently forwarded the message from ACM Queue on "post-relational" technology). UML is maturing, metadata is really picking up steam, and I believe ODMG is a much more solid specification than Date and others give it credit for. On a theoretical level, there are a lot of areas of work (such as ontologies and computational applications of modal logic) that I believe are very promising. At the time of Dr. Codd's seminal work, many of these ideas were only beginning to take shape. === Gregory Woodhouse [EMAIL PROTECTED] "Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away." -- Antoine de Saint-Exupery On Jun 11, 2005, at 1:48 PM, GARY MONGER wrote: From my perspective relational SQL databases have served well in a one size fits all sort of way, but they are hardly the summit of 21st century system design.  I don’t hold that 21st century necessarily holds a better solution either. I’ve seen some very well done raw MUMPS systems, even some with SQL mappings for those that like verbose queries.  SQL mapping of VistA data has been available for some time.  The HDR historical project will result in a SQL mapping of the bulk of VistA clinical data. A database, well designed and implemented, be it SQL, Fileman, MUMPS, filing cabinet, shoebox, or whatever can solve some problems well, others not so well.  No matter what the technology, no matter what the architecture, no matter what the design, any system that replaces VistA will have significant shortcomings.  The task is so massive, and so varied that no choice will serve every aspect well.  The real issue at hand is how to get it implemented well.  I think the VHA has and will continue to struggle with this.  It’s a tough problem, especially for a government agency. Managing implementation is also a tough problem for World VistA.  I can see that there is some good work going on, but I’m curious about how major efforts will be handled.  This data standardization issue is not an easy one to solve, it’s the nature of the clinical world.  Will that be tackled by World VistA?  What about name and number spaces?  How about mods to Kernel, HL7, and Broker?  I buy the open source model for development, at least theoretically, but we’re not talking about a Linux Kernel here.  In addition to the scale, VistA has very real ties to VHA development, which is likely to continue for quite some time, and now there is VistA-Office as well.   As for shooting VistA in the head ASAP, maybe we should get the new horse saddled up before shooting the one we’re riding.