Re: intermittent error "undefined near line..."

[Yuval Lifshitz]

thanks! this is indeed the bug. still have some questions though.

* the w_draws_b array is randomly initialized, but i thought that random values should be identical in any execution of octave (unless I seed them to time, which i don't)

 * also, the problem goes away quite consistently whenever i add an empty line anywhere in the file, and don't go away in any other case. could it be that octave is persisting its memory in some cases between runs?

Yuval

---

From: Dmitri A. Sergatskov <address@hidden>
To: Yuval Lifshitz <address@hidden>
Cc: "address@hidden" <address@hidden>
Sent: Monday, December 17, 2012 1:58:06 AM
Subject: Re: intermittent error "undefined near line..."

On Sun, Dec 16, 2012 at 3:23 AM, Yuval Lifshitz <address@hidden> wrote:

Hi All,

Running into the following error message every now and then. And was asked to bring that into the help list (see: http://savannah.gnu.org/bugs/?37918):

address@hidden:~/dev/economics/rania$ octave min_estimation.m 

GNU Octave, version 3.6.2

...

error: `ability_w_index' undefined near line 547 column 85

error: evaluating argument list element number 6

error: called from:

error:   /home/yuval/dev/economics/rania/estimation_f.m at line 547, column 35

error:   /home/yuval/dev/economics/rania/minestimation_test.m at line 5, column 6

error:   /home/yuval/dev/economics/rania/min_estimation.m at line 39, column 11

the variable is being set with a value at line 363 and there is no branching that i can see that can cause it not to be defined at line 547.

It is defined on line 363 only if "w_draws_b(draw_b,t,HS,1)<P_HUSBAND" condition is true.
Apparently it is not the case here. You can verify that by adding some disp() statements just after
that line.
 

Thanks,

Yuval

Regards,

Dmitri.
--