Re: Convert double to octave_value for assignment to column vector

[David Grundberg]

On 06/04/2010 03:49 PM, babelproofreader wrote:
> I still can't get this to work. My code now, heavily commented, is

<snip>

> which fails with cc:26: error: no match for ‘operator=’ in ‘retval =
> rm_compile_1_output’ and I don't understand why this should be so.

There are several mishaps with the types. An octave_value_list is 
distinct from an octave_value. One can't exchange them. It seems that 
octave_value_list and octave_value inherit each other somehow, which 
makes it trickier to catch. Also, feval returns an octave_value_list, 
not an octave_value.

> An alternative version is to uncomment the lines // *** and use my_retval as
> retval, which does compile without error, but when the function is called it
> produces
> error: octave_base_value::double_value (): wrong type argument `list'

Yes. You can't convert a octave_value_list ('list') to a double. You 
have to index the octave_value_list using the ()-operator. (Careful 
though, since the list can be empty!) That way you get an octave_value 
that you can run double_value() on.

> I am really at a loss to understand why, after converting input_median to a
> double and assigning it to the output column_vector in the loop, that this
> loop output does not seem to be recognised.

It is executed, it's just that Octave can't convert an octave_value_list 
to a double.

I've attached a changed file which runs and produces some kind of result.

Here to help,
David

rm_compile_1.cc
Description: Text Data