On Sun, 25 Apr 2010, Malcolm J. Currie wrote:
> Sounds correct without looking at the code. Just ignore the bad data points
> from the fit. There might be some historical reason why it is the way it is.
>
>> In fact doing this (just stopping the new point from being accepted):
>>
>> diff --git a/applications/kappa/kapsub/kps1_xyzw.gdr
>> b/applications/kappa/kapsub/kps1_xyzw.gdr
>> index 04283a9..3fdd34c 100644
>> --- a/applications/kappa/kapsub/kps1_xyzw.gdr
>> +++ b/applications/kappa/kapsub/kps1_xyzw.gdr
>> @@ -222,7 +222,7 @@
>>
>> * The variance is bad. Assume it is infinite and set the weight to
>> * zero.
>> - W( NGOOD ) = 0<CONST>
>> + NGOOD = NGOOD - 1
>> END IF
>>
>> * Keep track of the maximum weight so far.
>
> Have you committed the change?
>
No. I was waiting to see if you thought it was okay. It definitely fixes
the FITSURFACE problem.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|