[daip] ERROR: buffer size too small when applying BP table

[Eric Greisen]

Anne-Marie Brick wrote:
> Hi,
> I am trying to reduce some spectral line LBA data in AIPS, and I have 
> run into a problem when trying to apply the BP table to the data. I can 
> look at the data using POSSM, and I can look at the BP solution in 
> POSSM, but when I try to look at the data with the BP solution applied I 
> get the following error.
> 
> ITBSRT: BUFFER SIZE  199680 TOO SMALL FOR LREC= 32797
> BPASET: ERROR    1 SORTING BP TABLE
> 
> This also occurred when I tried to SPLIT a small section of the data out.
> 
> I have tried selecting a small time range of data (2mins) and a small 
> number of channels (the dataset is 2048 channels across 2 MHz), and I 
> keep getting the same error.
> 
> I tried to do a google search to see if I could find any information on 
> the error, but was unable to find anything.
> 
> Any help in resolving the issue would be much appreciated.

Your data set is at the max size allowed per record:
   2 pol * 4 IFs * 2048 channels = 16384
Apparently our standard buffer size - just adequate to read this is too 
small when overhead of a brute force sort is added (by ~100 words)

Any changes require massive updates of all of AIPS unfortunately.  The 
rollover to 31DEC09 is coming soon.  I will do 1 thing then: raise the 
limit to 32768 and raise the allowance for overhead as well.  I should 
--- and may have a chance to actually do --- change the sort to do a 
smarter sort in the case of long rows.  I am sure that the actual sort 
needed here is negligable save for the very long row size.

A work around might be to UVCOP the data set, dropping the bad spectral 
channels with BCHAN and ECHAN.  This should trim the size of the BP 
table and then sorts will work as currently coded.

Eric Greisen