[Babase] errors

Stephanie Combes babase@www.eco.princeton.edu
Mon, 12 Jul 2004 16:55:56 -0400


I'm adding more explanation/comments below (WITH ** in front of my new 
comments):

--On Monday, July 12, 2004 11:17 AM -0500 "Karl O. Pinc" <kop@meme.com> 
wrote:

>
> On 2004.07.12 09:06 Jeanne Altmann wrote:
>> Thanks Steph,
>> just a few notes at this point:
>>
>> At 05:38 PM 7/9/2004 -0400, you wrote:
>>> Hi again everyone,
>>>
>>> (Susan - I'm copying you on this, but if you don't want all of this
>>> info in the future, let me know).
>>>
>>>
>>>  I've looked through some of the errors (those that are listed on
>>> the first 40 pages or so).  Here's a list of ones that we need to
>>> discuss or that make me go "hmmm".  Email input is fine, or note
>>> whether certain ones need to be discussed on Wed.
>>>
>>
>>> 1) Biograph statdate can't be less than mature dates.  This happens
>>> for JUN and NIN.  We are changing this, correct? i.e. statdate won't
>>> really be the reference, but something that also checks their status
>>> (alive or not).
>>
>> RIGHT, THAT WAS AGREED ON LAST WEEK
>
> Right. Havnen't changed the rule yet.
>
>>
>>> 2) Rule violation: mature dates must be between 3 and 7 years of
>>> birth. This is violated for ALEX who matures around 7.5ish.  Leave
>>> rule and ignore Alex?
>>
>>
>> SUSAN, WHAT DO YOU THINK?  ALEX IS A TOTAL OUTLIER, AND WE DON'T HAVE
>> A GOOD BIRTHDATE FOR HIM AS HE WAS A JUVIE IMMIGRANT. COULD HE HAVE
>> BEEN SO MUCH YOUNGER THAN WE HAVE RECORDED?  DO WE WANT TO UP THIS TO
>> 7.5?  REMEMBER, WHATEVER AGE WE USE, THE PROGRAM WILL HOLD US TO IT,
>> WE CAN'T DO A 'BYPASS', RIGHT?
>
> Right.
>
>>
>>> 3) Cycles must be sequential for each female and the first seq for
>>> each female must be 1.  While these are sensible rules, it's messing
>>> up on a lot of females.  All of the females I checked (and I checked
>>> most of them) have seq's (in cycles) from 1 to whatever BUT there
>>> are many situations where their order in cycles goes something like
>>> 2,3,4,1,5,6,etc.  The conversion program doesn't seem to like this.
>>> Look at DAD or EST for examples, although there are others.
>> KARL, WHAT IS THIS ABOUT?
>
> When one of the cycles won't go in due to an error, all the rest with
> higher numbers won't go in either because that'd make a 'gap'.  Fix
> the one that won't go in and these errors go away.
>

**YES, BUT IF THE CYCLES ARE IN THERE AND THEIR SEQ NUMBERS CORRESPOND WITH 
CONSECUTIVE CYCLES DOES THE PROGRAM NOT PUT THEM IN?  I'M NOT SURE WHY THE 
PROGRAM IS HANGING UP ON THIS.  AS FAR AS I CAN TELL, THE DATA SHOULD GO IN.



>>
>>
>>> 4)  cycle is not a conception and has no dates.  This is ok, right?
>>> Apparently these are "fillers"? - because in some cases these lines
>>> are totally blank in cycles.  Of course, I am a little confused as
>>> to why there are totally blank lines in there in the first place.
>>> help.
>>
>> WE NEED FOR INFO ON THIS
>
> We talked about this.  This is for pregnancy conception cycles.  I thought
> I got the rule changed, but maybe I didn't or havn't installed it yet.
> Been busy with speeding things up.  Can Steph take a look and see
> if these are pregancy conception cycles?
>


**I CONFIRMED THAT ALL THE BLANK CYCLES IN CYCLES ARE CONCEPTIONS. SO 
DATA-WISE WE ARE OK.  KARL - I DON'T THINK THE RULE WAS CHANGED AND THE 
ERROR MESSAGE IT POPS UP IS WRONG.  OR MAYBE I GOT THAT MESSAGE BECAUSE 
PREGS ISN'T BEING CONVERTED YET???


>>
>>> 5)biograph statdate can't be less than cycles tdate or ddate.
>>> Similar to error 1 above.  again - status is being added as a
>>> reference correct?  of course, this particular kind of error did
>>> help me catch one female who apparently cycled after she died.  oops.
>>
>> RIGHT
>
> Right. See above.
>
>>
>>> 6) sexskins dates overlap with the following cycles M date.  Yes,
>>> does happen for LULU (see LUL seq 43).  help jeanne.  I'm thinking
>>> that the M date should have gone with the previous cycle? or is this
>>> rule not quite fine-tuned enough?
>>
>> I'M NOT CLEAR ON THIS
>
> Check to see if it's a data problem or a rule problem.
>

**I THINK THIS IS A LULU PROBLEM (POOR, ODD BABOON).  JEANNE - THE ERROR IS 
POPPING UP FOR LUL'S CYCLES IN JULY 1976.  SHE DETURGESES BUT NEVER GOES 
FLAT.  SHE THEN STARTS GOING UP IN SIZE AGAIN (AND MY GOD - DID SHE REALLY 
GO UP TO SIZE 20?!?!? HOW DID THE POOR THING SIT DOWN???) WHAT'S HAPPENING 
IS THAT IN SEXSKINS, THE SAME CID (927) IS USED FOR WHAT'S BEING LISTED AS 
2 DIFFERENT CYCLES IN CYCLES (THE MDATE FOR THE NEXT CYCLE IS FALLING IN 
UNDER WHEN SHE WAS SUPPOSEDLY SWELLING FROM THE PREVIOUS CYCLE).  I THINK I 
NEED TO CHANGE HER INFO IN SEXSKINS.  BUT, JEANNE, DO YOU WANT TO CONFIRM?

>>
>>> 7) error: sexskins cid blah: maturedates.matured cannot be less than
>>> sexskins.date.  OK, in theory this error makes sense to me.  But,
>>> the female it's referring to matures well before the listed sexskin
>>> dates.  So karl, where is it looking?  what's maturedates.matured?
>>
>> KARL?
>
> Maturedates is now on the MATURES table, not biograph.
>
> I think this rule is just wrong and it should be the other way around.
> Right?
>
>

**WHERE IS THIS MYSTERIOUS MATURES TABLE?  SHOULD I HAVE A COPY OF IT 
SOMEWHERE OR IS IT A NEW CREATION?  I THINK THE RULE IS BACKWARDS.


>>
>>> 8) cid on cyles - key referenced from sexskins not found in cycles.
>>> scary message.  couldn't find source of error.  I'm still trying to
>>> figure out which female this is referring to.
>
> Doesn't it give you the bad cid on sexskins in the message?  Very likely
> this is a cycle row that didn't make it into the new database because of
> another error.
>

**NO, IT DOESN'T GIVE ME THE CID FOR THIS ONE. BUT IT MAKES SENSE THAT IT 
IS LIKELY A DOWNSTREAM ERROR.  FIXING THE OTHERS MAY FIX THIS ONE.


>
> Karl <kop@meme.com>
> Free Software:  "You don't pay back, you pay forward."
>                   -- Robert A. Heinlein
> _______________________________________________
> Babase mailing list
> Babase@www.eco.princeton.edu
> http://www.eco.princeton.edu/mailman/listinfo/babase