[Babase] Dupe Cenids in DEMOG
Karl O. Pinc
babase@www.eco.princeton.edu
Thu, 25 Aug 2005 00:32:18 +0000
On 08/24/2005 06:51:31 PM, Catherine Markham (amarkham@Princeton.EDU)
wrote:
> >
> > Anyhow, the new system clearly catches the error or it wouldn't
> > have spit out a problem during the conversion.
>
> The new system is catching the error in the conversion, but that
> doesn't necessarily mean it'll catch the error if I enter funky data
> in some future demography update, right? That's more my issue for the
> big picture -
It's supposed to mean exactly that, that it will catch the error
no matter how it's entered into the database. This particular
error I'm confident it will always catch. (Some others are
trickier -- mostly when you can get the same error state by
different paths of change in the data, particularly when
multiple tables are involved.) The conversion does not do
any error checking. It just inserts the data into the database
in an order that does not violate any rules, biograph first, etc.
I could have skipped all the rule checking as the conversion
loads the data, but then we wouldn't be checking either the data or
my rules. The rules are built into the database, so we can use
any program or none to alter the data and be assured the data is good.
The only tricky part is that there's more than one way to get the
data into the same state. First change this table then change that,
or vice versa, for example. As the conversion is loading every thing
the same way the rules are not fully exercised and no doubt there will
still be bugs that allow bad data. But it's better than what we
had before.
Karl <kop@meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein