Re: PvWave: Undefined variable ON_ERR_ACTION [message #17937] |
Fri, 19 November 1999 00:00 |
Adrian Clark
Messages: 6 Registered: October 1999
|
Junior Member |
|
|
Dear Andreas,
Your problem is due to an uninitialised COMMON BLOCK statement, I believe.
Take nonlinregress.pro, for example. The procedure for this can be found
in $VNI_DIR/stat-2.0/lib/nonlinregress.pro. The variable on_error_action
is defined as a COMMON variable and then you have the line:
ON_ERROR, on_error_action
In this case, the on_err_action variable tells wave what to do when there
is an error (e.g. return to the $main$ level, return to the calling
function level etc). However, the common block statement has to be
defined, otherwise you will get the error you describe, since wave does not
know what value to assign to on_err_action.
This COMMON block is initialised when you start the maths/stats functions.
i.e. depending on your setup, you will have to do one of the following:
wave> @stat_startup
initialises stats package.
wave> @math_startup
initialises maths package
wave> @ip_startup
initialises image processing toolkit (which includes maths & stats I think)
etc etc
I you haven't done this, it will probably solve your problem. If you have,
let me know and I'll look into it further.
Good luck!
Adrian
Andreas Liehr <obi@uni-muenster.de> wrote in article
<38341EAA.325C8271@uni-muenster.de>...
> Hi Everybody,
>
> I`ve observed a problem using the IMSL Mathematics und Statistics
> package. The problem occurs with the PvWave versions of VMS, Unix and NT
> while calling functions like NONLINREGRESS, NONLINOPT or ODE. These
> functions also don`t work with the examples of the help file.
>
> By calling one of these functions, I get the error message:
>
> % ON_ERROR: Variable is undefined: ON_ERR_ACTION (CMAST$GEN_COM).
> % Execution halted at NONLINOPT <nonlinopt.pro( 29)> (ON_ERROR).
> % Called from $MAIN$ .
>
> I haven`t found a documentation of this error message or the variable
> ON_ERR_ACTION anywhere and I`m hoping that anybody submits a hint or a
> solution to avoid the descriped error.
>
> Thanks,
> Andreas
>
|
|
|