[Babase] biograph update_statdate and dcause questions
Karl O. Pinc
kop at meme.com
Thu Jan 3 12:04:59 EST 2008
On 01/03/2008 09:41:48 AM, kfenn wrote:
>
> Karl O. Pinc wrote:
>> The system ensures that statdate is the last census date whenever
>> the individual is alive. You don't need to check/update new
>> individuals any more than you need to check/update existing
>> individuals' statdates. We tested this thoroughly, but you're
>> welcome to test some more.
>>
> I'm checking everything right now to make sure what I THINK I'm doing
> is actually showing up in Babase. I trust your testing, I just don't
> trust me to always interface with your system correctly.
It's been a long time testing, but surely there's bugs yet to be
discovered!
I thought
> that if I could leave the statdate blank for new individuals, then
> having it populated after the census uploads would confirm for me
> that I set things up correctly for the update to occur....and that I
> actually loaded census info for that individual so that the statdate
> was updated. I guess putting a birthdate into the statdate field
> just to get it uploaded is OK, but the birthdate really isn't the
> statdate (we rarely see an individual on the day they are born) and I
> wouldn't ever want it to get left in there accidentally. I'm more
> inclined to input some really fake date to get around the "cannot be
> null" issue....like 2050-01-01, something allowable, but so out of
> whack that it would catch anyone's attention and I could query by it
> to make sure it was removed. Is that a bad idea?
Not a bad idea. Thinking it through though might raise issues.
The system's designed to prevent you from putting in
unreasonable data. But it does not test everything and I
think what you suggest would work.
The goal would be to come up with some unreasonable data that's
noticeable. Maybe birthdate=statdate and status = alive is enough,
because after a while at least if they're still alive then you'd
expect them to have a more recent statdate. I was thinking to just
look for living individuals with old statdates, but those pesky
males get in the way. Of course the statdate - birthdate computed
age would be small and that could be added to the test. But
after a while things get complicated and un-obvious.
Maybe your idea is the best.
But what sort of errors would be caught? Seems to me the only
problem would be if the new individual's Sname as entered
in BIOGRAPH is different from that in CENSUS. Unless
the Sname going into CENSUS is already in BIOGRAPH you'd
get an error trying to update CENSUS. If there's nothing
in CENSUS then the individual should go in dead with the
right Statdate anyway. So there's 2 errors: 1) Using the wrong,
and a pre-existing, Sname in the CENSUS table. Basically
this means getting the Sname wrong forever. _Eventually_
this would, probably, cause some sort of conflict.
2) Entering the wrong Sname in BIOGRAPH, and then
entering a new row with the right Sname but not removing
the bad BIOGRAPH row.
Then there's probably another way to make bad things
happen I've not thought of. Oh well....
Karl <kop at meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein
More information about the Babase
mailing list