[daip] Indxr changes in 31DEC14?

[Eric Greisen]

On 06/06/2014 07:40 AM, Arnaud Collioud wrote:
> Eric, Amy,
>
> I am processing a VLBI experiment with an MNJ-up-to-date 31DEC14.
> I need to regenerate CL1 using INDRX. When running the task on my Mac (10.6.8) laptop, it stops suddenly with the following message:
>    macpco> INDXR1: Task INDXR  (release of 31DEC14) begins
>    macpco> INDXR1: Creating CL table with version number 1
>    macpco> INDXR1: ZABORS: signal 10 received
>    macpco> INDXR1: ABORT!
>
> Loading the same data on another machine (Linux 32bits) with 31DEC13 installed on it, then executing INDXR works fine. Did you change something in INDXR (or related functions) on 31DEC14?

I checked the revised subroutines and there seems no
zero divides that were not in the old code - but I
just defended against them anyway.

Signal 10 on Macs means "bus error" - usually a bad address.
I don't see anything in the changed subroutines that is not
defended properly.  That doesn't mean there isn't something
just I can't see it.  Perhaps, of course, there is a compiler
issue.  I will need proper data to test this perhaps Amy has some
that I could try before requiring yours.

Cheers

Eric Greisen