Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 19 Jul 2010 15:39:46 -0500

Hi Lewis,

On Friday 16 July 2010 17:11:22 J. Lewis Muir wrote:
> MEDM crashes if I open scan_full.adl, execute an sscan scan which
> produces large detector values for the first detector, wait for it to
> complete, and then open scanDetPlot.adl by clicking on the 01 button in
> the Detectors section and choosing "Plot Array".  Attached is a GDB
> stack trace in case it is useful.

Your line numbers don't seem to match mine; line 2115 is not in LineSet(), 
although it does execute that line when LineSet() calls ItemGetNew() and the 
the items further down the stack are consistent with that line number being 
correct.  This may be due to optimizations by your compiler, so I'm not too 
concerned by it.

> Maybe this has something to do w/ how the large double gets converted
> into a float??

The crash is probably related to the two NaN values that are shown in your 
stack trace as arguments to LineSet(), which I suspect come from NaN value(s) 
in the dataset; Infinities and NaNs can also be generated by CALC expressions 
in recent versions of Base, or can come from numeric conversions or other math 
functions.

I have not attempted to reproduce the crash myself.

- Andrew
-- 
The best FOSS code is written to be read by other humans -- Harald Welte

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System