Jim Self wrote:
> >- SQL(92) compatibility (Interbase 6.0, (PostgreSQL 7.0 close to))
>
> Although I agree that some query language is necessary I don't think that
> SQL is all that desireable for searching patient records. Although I was
> aware of SQL and had studied it as a graduate student in computer science
> when I started working on VMACS almost 20 years ago, I have almost totally
> avoided it since then.
I agree. Specialist query languages at the application level are more likely to
be useful. Whether the datbase is doing SQL under the cover or not may not be
important. And in any case, it would be better to use OQL or SQL3, so as to have
object semantics.
> One area of concern that I think is critical to healthcare oriented systems
> yet outside the realm of many RDBMS is the ability to index and retrieve
> words within large bodies of text. In looking through the documentation on
> Postgres at http://www.postgresql.org/docs/ I was not able to find any
> mention of how or whether this could be done. I did find a cryptic reference
> in http://www.postgresql.org/docs/admin/release.htm to 'sentence.word' which
> suggests that there may be provision for this and that perhaps it is only
> the documentaton (or my reading of it) which is lacking.
"Content based retrieval", or CBR. This would most likely be done these days
with a meta-model approach; future databases appear likely to employ XML-query
for this.
> In the context of HTTP/HTML, at least, it is not at all necessary that large
> objects, such as radiographic images, be stored directly within a database.
> They could as well be stored simply as image files and served separately
> provided that their location and other characteristics are recorded in the
> database.
Agree.
- thomas beale
