[Babase] Fwd: babase update and conf call

Susan Alberts alberts at duke.edu
Tue Sep 12 17:25:34 EDT 2006


Sorry, I don't know why I didn't send that the mailing list! Here is  
it again.

Susan

Begin forwarded message:

> From: Susan Alberts <alberts at duke.edu>
> Date: September 12, 2006 5:24:13 PM EDT
> To: "Karl O. Pinc" <kop at meme.com>, Jeanne Altmann  
> <altj at Princeton.EDU>, KETHERINE FENN <kfenn at Princeton.EDU>, Leah  
> Gerber <lgerber at duke.edu>, Catherine_Markham Markham  
> <amarkham at princeton.edu>
> Subject: babase update and conf call
>
> Dear all,
>
> Karl is about to reach a solid end point and we are excited about  
> this. There are some loose ends that will need to be tied up, and  
> we think that a babase conference call for Thursday would be a good  
> thing. We would really love Catherine to be involved if at all  
> possible becuase of her relatively long history with babase (and as  
> a future user), but I realize her schedule is now quite different.
>
> Jeanne and I have a different conf call at 11. We could do our  
> babase conf call before that, at 10 am, or we could do it bewtween  
> 2:30 and 3:30. That is about all the time that I have free that  
> day. If that can't work out then let me know and we will discuss  
> another time after Karl is back in Chi, but it seems like it would  
> be good to do it before he leaves here (but not absolutely necessary).
>
> Here is the update and agenda for our call:
>
> Here are the things that still need to be done before Karl leaves.
>
> Cycgaps validation and assigning cycles.series AND automatic  
> cycgaps updateing. This is in progress now.
> Reconversion using the new automatic updates.
> Update Ranks documentation (is this the last table for which the  
> documentation reflects the old foxpro system?)
>
> At that point, the conversion is done, except for then debugging to  
> deal with conversion errors that are programming errors (if any).
>
> After this final conversion, there are some things that we will  
> need to do and decisions we need to make that affect how we use the  
> database (I am NOT including in this list the things that we all  
> know are outstanding, such a developing the data entry procedures,  
> adding new tables, creating new ranker, and fixing the front end).
>
> Non-trigger error checks. There are a set of error checks that we  
> can run to check the integrity and accuracy of the database. The  
> errors that this would find are not errors that could or should be  
> coded as automatic triggers, hence we will describe these as non- 
> trigger error checks. They are things like checking that every  
> animal present in the group has a rank assigned (as opposed to  
> checking that no one is assigned a rank unless they are in the  
> group – this latter one is a trigger, the former cannot be because  
> before ranks are assigned, no one has a rank).  Karl notes that we  
> have always had these non-trigger error checks available to run in  
> foxpro, but that no one has every really run them or used them to  
> identify and correct errors. Hence, they are effectively optional  
> error checks. The code has been written for these error checks, but  
> not debugged and hence not run before. Our options are:
> Take the error checks that Karl has written and debug and run them  
> ourselves, one by one as needed. They are sql queries, so this is  
> highly feasible but laborious.
> Have Karl debug and concatenate them into a single program that  
> would be run whenever we wanted it to be, probably automatically on  
> a regular basis.
> Outsource this project to someone else – i.e., hire someone to  
> concatenate and debug these programs into a single program that  
> would be run automatically.
> Testing and pushing the system. Karl’s concern as a professional  
> programmer is that we find the worst bugs early on and fix them. He  
> points out that of course there will be bugs. To find them, we need  
> to push the system – do a lot of manipulation of the tables using  
> babase_test. Update rows, delete rows, make illegal and legal  
> changes and see how the system performs. Ideally this should be  
> done for all tables, all columns. The sexual cycles tables should  
> have special attention here because there are so many “analyzed”  
> tables. This kind of testing will also reveal performance issues  
> that might come up that we haven’t anticipated or found yet.  One  
> approach would be to just devote a certain amount of time in Leah’s  
> and Tabby’s week, some week, to doing this. Another option would be  
> contracting Karl or someone else to do this. It might be a couple  
> of days work, hard for me to say.
> Dealing with warnings and unwanted error messages. Every night, the  
> program will run the triggers, and these automatic checks will  
> produce a list of error messages and warnings related to data  
> “errors” in the database. Specifically, anytime a rule is violated  
> (rules outlined in the documentation) a message will appear. Some  
> of these we will want to ignore forever or for the time being, and  
> we will not want to be revisited with them every day. So, Karl  
> proposes a program that will defer or eliminate unwanted error  
> messages and warnings. This is optional, but would be a great  
> convenience. It is not really something that I think is feasible  
> for us to do without professional help, unless Leah or Tabby have  
> insight into this. What to do?
> Checking cycle ids? Karl points out that in creating the new cycle  
> files in the new babase system, cycles have different cids in many  
> cases than they did in the old babase system. He wants to know  
> whether we want a program that maps old onto new, so that we can  
> find particular cycles that we analyzed in foxpro.
>
> -----------------------------------------------
> Susan Alberts, Dept. Biology, Duke University, Durham NC 27708.  
> Phone 919-660-7272, Fax 919-660-7293. alberts at duke.edu
>
>
>

-----------------------------------------------
Susan Alberts, Dept. Biology, Duke University, Durham NC 27708. Phone  
919-660-7272, Fax 919-660-7293. alberts at duke.edu



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.eco.princeton.edu/pipermail/babase/attachments/20060912/2cd10447/attachment.html


More information about the Babase mailing list