[Babaseannounce] Possible (fixed) problem with PSION data,
and other possible (ongoing) issues
Karl O. Pinc
kop at meme.com
Sun May 6 23:27:35 EDT 2007
Hi,
It may be that all of the most recent PSION data did not
get copied from the foxpro database into the babase
database, when the data was copied a few days ago.
The SAMPLES.Setupid values in some of the most
recent PSION data were bad, and Leah just corrected
that data today. None of the "bad" SAMPLES, or related
POINTS, NEIGHBORS, etc would have transferred
from foxpro into the new system, although it would
have been in foxpro (with a bad Setupid.) The old data is gone
now, as is the record of what did not copy,
so I can't say for sure what happened. (If
I had to I could figure it out.) Because
we had a lot of problems with the RANKS data, data
that was bad in foxpro and so result in "bad"
data in the new babase database, I didn't take a close look
for other problems involved in moving the data
into the babase database.
I hope this has not caused anybody problems and apologize
if it did.
We are overwriting the babase database now with new
foxpro data. I expect the new data to be fine
and the babase database to once again be an exact
copy of the foxpro data. I'll know for sure when
the conversion finishes and I review it's logs.
At that time I'll let you know if there's any problem
with the "live" data.
The problem occurs because not _all_ checks are turned
off when copying the foxpro data directly into the
babase database. Certain fundamental checks
remain, things like checking that an Sname is valid.
We've not had any of these sorts of errors in a long time,
but some have cropped up in the new 2006 data. Come to think
of it, I seem to recall at least one error on the NEIGHBORS
table having to do with an invalid Sname, so it may be
that what's in the babase database will not be exactly
what's in foxpro. (Whether it really matters or not
is another issue. I suppose that depends on the exact
error and we won't know about that until the
conversion finishes later tonight.)
(Lacey and Tabby,
I'm putting the errors that show up while loading the
"live" babase database in:
/biology/groups/babase/errors/errors_live_v14
You can look at them there and we can decide whether
another posting to this list is called for.
It contains the output of the programs that rebuild
the "analyzed" data, things like MEMBERS, CYCGAP_DAYS,
automatic Mdates, etc. so expect to see extra output
showing up after the "usual" conversion error stuff.)
Point to take away: I expect the pre-2006 data is fine,
just as it has been. The 2006 data _should_ be the only
area of concern because it's the only area where we'd
introduce problems severe enough to show up while loading
the babase database.
If there are any questions or if you'd like any clarification
please write the babase at eeblistserv.princeton.edu email
list and we'll have the discussion there. If necessary
we'll then send a summary of the discussion back to this list.
Karl <kop at meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein
More information about the Babaseannounce
mailing list