NYCPHP Meetup

[Brian D.]

> Second, there were ideas that came out during Q & A that might be
> expanded on to good purpose.  Things that we might not know or take for
> granted, like the philosophy regarding types and structures. The
> audience seemed to be focusing on the need for structure, while the tool
> did not seem to want to do that.

That made me wonder if most people completely missed the point. The
application of XML databases is, I think, in situations where
structure is either not applicable or not possible. Trying to stamp a
structure on an XML database (from what I can gather) destroys one of
the primary reasons for employing the technology. XML is flexible.
That's what makes it different. If you shoehorn an XML database into
what Rusty called a "rectangular" format, why not just continue using
relational databases?

As far as uses go, I was nodding the entire time that Rusty was
talking about the usefulness of structure-free systems in the medical
industry. I've worked in the medical industry almost exclusively for
seven years and there have been several times that I have had to force
data (documents, specifically) into relational formats that obviously
didn't work well. As Rusty pointed out, it makes a developer feel
dirty. I've seen "documents" tables with over two hundred columns, and
it took all sorts of work-arounds to fix the performance problems that
it caused. I've also seen the EAV format that Ali M. mentions in his
email. None of these solutions seem to fit the problem very well.

Rusty's presentation interested me because it seems to hold the best
answer I've seen yet to the free-flowing/constantly-changing data that
medical software has to deal with. I still don't feel that the (open
source, anyway) XML databases are a *mature* solution, but they
certainly hold a lot of promise for an industry afflicted with poor
containers to hold our data. The hybrid solutions hold the most
possibility, I think, because it meets the need for both structural
and non-structural data. I will be watching the maturation of the
available software with much more attention than I have before.

- Brian Dailey


realm3 web applications [realm3.com]
freelance consulting, application development
(423) 506-0349