[daip] VBGLU bug

[Cormac Reynolds]

On Fri, 27 Jan 2006, Eric Greisen wrote:

> Now I am confused.  The algorithm involves getting all the data
> records for a particular time into BUFALL (called BUFOUT in lower
> level routines).  OUTSIZ is set at the top as a parameter
>       PARAMETER (OUTSIZ = 128 * 16 * (190 + 20) * 3)
> or 1290290.  The size needed should be
>       OUTSIZ = (3 * NSTOKS * NCHAN * NIFOut + NRPRMO) *
>      *   ((ANTMAX * (ANTMAX+1)) / 2)
> or in your case
>       outsiz = (3 * 4 * 512 * 2 + 9) * (8 * 9) / 2
>              = 12297 * 36  = 442692
> Even if there is a problem and it needs 2 of these there should still
> be more than enough space.
>
> I was going to put in dynamic memory to deal with this, but apparently
> there is some non-obvious algorithm problem...
>
> I may need to fetch the files but perhaps we can avoid that or fetch
> files containing only a few times.  Does the program abort early or
> does it get rather well along?

It takes quite a while to reach the abort, and I am suspicious that it takes a
large dataset to cause an abort (I have seen this happen twice now and both
times the datasets were of order 10 GB). I can easily make the files
available and the internet connection between here and Socorro seems pretty
good. Just let me know if you want them.

cheers,
Cormac.