[Babase] six month plan

Susan Alberts babase@www.eco.princeton.edu
Thu, 6 May 2004 16:49:00 -0400


Hi all,

This is a summary of the conversation that Jeanne, Karl, Daphne and I 
had yesterday May 5. Please read this and make sure I summarized 
correctly, and confirm that we all have the same understanding. 
Steph, a lot of this pertains to you even though you couldn't be at 
the meeting yesterday, so let us know if you have concerns or 
questions.

1.Karl has made good progress and has started doing data conversions. 
He will continue doing data conversion, data integrity checks  and 
program conversions. Originally we thought these tasks would be sort 
of sequential but in fact they are somewhat simultaneous at this 
point.

2. As the progress goe on, errors in babase will contine to emerge. 
Some will be more important than others to correct quickly. These are 
errors that could potentially cascade, i.e., produce errors in a 
series of files "downstream" from the original error. Other errors 
(e.g., the ones in repstats that Steph identified) will themsevles be 
"downstream" errors with few implications for overall data integrity.

3. The ideal would be to correct all errors as they come up. However 
this is probably not realistic.

4. So, we all agreed that we need to carefully scrutinize errors to 
determine whether they are "upstream" errors with the potential to 
cascade and mess other things up, or "downstream" errors that can 
wait to be fixed. We need to make upstream errors a very high 
priority for correction.

5. Karl will be key in the process of identifying upstream versus 
downstream errors. However, we all will need to think carefully about 
the errors that come up, as prioritizing errors will be a team effort.

6. Steph will be the main point person for running data conversions 
for Karl when he says it's time, and getting the lists of errors that 
result each time. Karl has already instructed Steph to get a login on 
the Duke server. He will also step her through the conversion process.

7. Steph will also be the main point person for fixing errors, with 
help from Daphne when it makes sense. However, she will need lots of 
help in prioritizing these corrections (see points 3-5 above). We 
really all need to work as a team here, especially to make sure that 
Karl is able to stay on task rather than having to wait on the 
correction of upstream errors.

8. Steph sill send an email to the babase mailing list each time she 
finds errors, and that's how the rest of us will work as a team to 
talk about prioritizing error correction.

9. Susan will be out of touch while in Amboseli, June - mid-August. 
By mid-August she'll be in Nbi and in good email contact (hopefully).

10. The US team (Karl, Jeanne, Steph, Daphne) will have a phone call 
each week, tentatively Weds at 1 pm (did I get this right? You guys 
can modify as needed).

11.  Because Steph is working part time until sometime in July, 
she'll need to send Karl her work schedule in Susan's lab, to keep 
him continually updated on her availability to him. Starting some 
time in July she will be full time in Susan's lab.

12. We need to keep a running list of all errors and whether we have 
fixed them, and I think it can't just be the babase mailing list. 
Steph, you and I need to come up with a system for this, and for 
indicating how we prioritized the errors and what their status is. 
Can you put this on your list of things we need to talk about please? 
Anyone else want to give input on this?

13. We identified a general problem that Karl will deal with. 
Sometimes the babase team might want to make changes in  a coding 
system, as happened with dcauses last year. Karl's babase programs 
and data integrity checks make assumptions about the coding systems. 
He will therefore disallow changes in coding systems that affect data 
integrity checks. To change coding systems, we will have to discuss 
the change and get it implemented throughout the system.

I think that's it for the 6-m plan.  A few other points.

-We need to get the babase documentation updated to reflect the new 
dcauses. Currently the old dcause list is in the babase manual. We 
could do this here, but we will need the descriptions that you came 
up with for the new dcauses.

-Karl has made a suggestion that we give all dead animals a dcause 
(currently all prior to mid-2001 have "no dcause"). I think this 
needs to be done at Princeton, as we have an informal policy (which I 
think we should adhere to rigidly) that we won't make permanent 
changes to the database at Duke until the new system is up and 
running.  If we start making some changes at Duke at this point we 
run a real risk of losing track of which copy is the "real" copy. 
Especially if the change takes days or weeks, and some changes are 
being made to the same file at Princeton, even inadvertently. So, 
Jeanne and Daphne should come up with a plan for doing this.

Sorry for the long email, I think that's it for now.

Susan


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Susan Alberts, Assistant Professor
Department of Biology, Duke University, Box 90338, Durham NC 27708
phone 919-660-7272  fax 919-660-7293