Allow non-constant value for vector I[#value,index] instead of just i[#value,index]

[Reptorian1125]

Right now, I have this code:

		modifiable_array_ind=$start_img_ind+(t<<1);
		unmodifiable_array_ind=modifiable_array_ind+1;
		current_coordinates=I[#modifiable_array_ind,0];

This leads to:

[gmic] *** Error in ./+rep_random_subdivision_of_dimension_boundaries/ (file 'C:\Users\Miguel\user.gmic', line #67) *** Command 'eval': [instance(1,1,1,1,000002AF10CD46D0,non-shared)] gmic<float32>::[gmic_math_parser] gmic<float32>::eval(): Value accessor '[]': Specified image index is not a constant, in expression 'current_coordinates=I[#modifiable_array_ind,0]'.

A non-constant value will work for i[#value,index] though.

Or even just different consts per threads defined by begin_t().

[Reptorian1125]

My solution is this:

modifiable_array_ind=$start_img_ind+(t<<1);	
unmodifiable_array_ind=modifiable_array_ind+1;
copy(current_selection[0],i[#modifiable_array_ind,v(0,da_size(#modifiable_array_ind)-1)],coordinates_and_iterations_array_size,1,h(#modifiable_array_ind));
da_remove(#modifiable_array_ind,0);

This allows me to use non-constant value without using I[#value,index]. Kinda unweldly, but it is what it is.

[dtschump]

Yes, using copy is the right way.
The fact is that when you specify a non constant image index (so potentially variable), the math compiler cannot determine the spectrum of the associated image at compile time, and thus the size of the output vector.
As vector objects are static, their sizes must be known at compile time.