memory errors on iterorigin

[Danny Jacobs]

Hi all,
Have any of you seen memory errors when using the measurement set
object iterator methods?

It usually manifests as either a memory leak (memory use grows larger
than entire data set size and apparently without bound) or a
"std::bad_alloc" error. This usually happens when doing
ms.iterorigin(). I've not been able to recreate the error on any NRAO
machines, so until I can get more data it will be hard to track down.
The attached script triggers the problem every time. Is anyone else
using this version of linux and successfully iterating?

Thanks,
~Danny

System details:
CentOS 64b linux v2.6.18
Also seen on OS 10.5, 64b
casa 3.2.0 r13717

--
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Danny Jacobs

Dept. of Physics and Astronomy
University of Pennsylvania
209 South 33rd Street
Philadelphia, PA 19104-6396
-------------------------------------------------
Office:        Rittenhouse Lab - 1n2a
Phone:        (215) 280-7357
Fax:        (215) 898-2010
Homepage:     http://dannycjacobs.googlepages.com

[Urvashi]

Hi Danny,

I see this problem too. Getting it right now. Will poke at it.

Urvashi
(उर्वशी)

[Urvashi]

Hi Danny,

Reason found, and being fixed in 'active'.

To avoid this problem on an old build, set "interval" and
"maxrows" to 0 in the call to ms.iterinit(). Defaults were not being
set automatically. Kumar has updated all the defaults, and we're
checking that they work....

Urvashi
(उर्वशी)

On Mon, Apr 25, 2011 at 1:34 PM, Danny Jacobs <jaco...@sas.upenn.edu> wrote:

[Danny Jacobs]

Supposedly it was fixed in "active" 3.2 r14492 around the beginning of
the month.

However I have still seen it occasionally. Its really hard to
replicate, but I could believe the defaults problem. It usually
happens at the very beginning of a session before the ms module has
been used.

Thanks for the update!
~D