[Babase] DIFFS output
Karl O. Pinc
kop at meme.com
Mon Jun 11 10:13:44 EDT 2007
On 06/11/2007 08:15:51 AM, Lacey Maryott wrote:
> Karl,
> I am getting really confused about the output into members.diff...
> I am looking at it as if the (-) is what is in the old data and the
> (+) represents the new data I just put in... however... toward the
> end of the file, there are 1.20's and 1.40's in (+) columns... and I
> am pretty sure I didn't put anybody in anything except 1.00. Also,
> there are animals in 1.40 before the groups statdate in 1989... :-\ I
> only deleted census records for grp 1.00, but for example, one
> individual who is demonstrating this is...
>
> - FAV | 1988-12-27 | 1.00
>
> - FAV | 1988-12-28 | 1.00
>
> - FAV | 1988-12-29 | 1.00
>
> - FAV | 1988-12-30 | 1.00
>
> - FAV | 1988-12-31 | 1.00
>
> + FAV | 1988-12-27 | 1.40
>
> + FAV | 1988-12-28 | 1.40
>
> + FAV | 1988-12-29 | 1.40
>
> + FAV | 1988-12-30 | 1.40
>
> + FAV | 1988-12-31 | 1.40
>
> The last census date for 1.00 in that month is 1988-12-21... is there
> some reason it is shoving him in 1.40 because of interp rules? He is
> still in the census in January, but that was the point at which the
> groups were switched from 1.00 (which ended 1988-12-31) to 1.40
> (starting 1989-01-01). So, should I assume things like this, at
> this time period, are just expected diffs because of interp?
If FAV is last censused in 1.0 on 1988-12-21, then not censused
again until appearing in 1.4 in the beginning of Jan 89 then
interpolation assumes FAV stayed in 1.0 for a while and then
switched to 1.4. Whether this is "right" depends on what the
scientists want. There was a lot of arbitrary deciding who's
where and what census was of what group when. If you need to
enforce those decisions then "N" (or "M") Status CENSUS rows
must be put in to tell interpolation who should be where.
No one really wants to think about this but the system
has (and always has had) support for censusing of sub-groups.
So long as all the effort is going in to re-entering who was
censused where it'd be worth deciding once and for all
how much advantage is to be taken of this feature. (Recall
I can also add the supergroup() function as a column in
a view of MEMBERS to make sub-group censusing completely
transparent when querying. That will take as long as it
takes me to type the documentation.)
You may want to ask Jeanne about forwarding this question
to Susan.
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