[MR] RE: heraldry

[Michael Houghton]

Howdy!

On Thu, Jun 07, 2001 at 08:48:27AM -0700, Christofe DuBois wrote:
> Databases are nothing more than delimited files.  With very little / no
> additional work, I can open most database files on most other database
> programs.  The data will port over to new database solutions just fine.

I will be blunt. That statement is so general as to be meaningless when
applied to the Armorial database.

Before proposing grand designs to rehost a database, it behooves one to
understand the existing system. Without that understanding, the design
is almost certain to fail to address the particular nuances of the system.
For the Armorial, see http://sca.org/heraldry/OandA/data_format.html
This gives the first order data one needs to interpret the data. It does
not speak to interpreting the indexing data.

Having given thought from time to time to putting together a relational
schema for the data, I can say that it is not as easy as it may appear.
Remember that there is historical data retained in the database. 
Remember that the name can change. Note that mistyped names in the LoAR
get recorded (and maybe corrected explicitly later). Items get reblazoned
or corrected. 

The present Armorial format represents a highly denormalized database and
actually does not include all of the data one needs for the full database.
The list of categories and features and cross-references is maintained
in a separate file. Ancillary data further complicates the picture, as
one needs to know how to break the data up for the categories in the 
Ordinary. After a bit more than a year in the job, I am still learning.
I am very grateful to my predecessor for handing me a well oiled machine
that I did not have to create myself, and for being available to answer
my questions as I came up to speed. 

It really is not as simple as Christofe would make it sound.

Further, when proposing a "real-time" sort of system to replace the 
batched system, one must consider how that change will actually make
the system work better. One must consider the nature of the workforce
that makes it happen, and the constraints imposed by the need to have
a deliberate process (in the sense of considered discussion and 
thought). Moving to a "real-time"-ish approach would tend to encourage
a hasty process. This requires further understanding of the "business"
objectives and process.
> 
> As far as glitches go...  I have a one word response-  Redundancy.  It's
> still an infinitely more reliable source than paper, which can crumple,
> wrinkle, rip, blur, smudge, and, as we started this thread with, fade.
> 
That depends on the use. For conflict checking, a paper copy is not 
dependent on electrical current (and is usually of comparable speed to
electronic searches). Ultimately, the paper LoARs are the definitive source
for the information. 

yours,
Herveus
-- 
Michael and MJ Houghton   | Herveus d'Ormonde and Megan O'Donnelly
herveus at radix.net         | White Wolf and the Phoenix
Bowie, MD, USA            | Tablet and Inkle bands, and other stuff
                          | http://www.radix.net/~herveus/