Jesse

I'm not understanding something about jit.expr and could use some help.

In the attached patcher I'm applying a variable scaling via jit.rota with a fixed anchor point. I want to apply this to plane 0 of a three plane float32 matrix.

Panel A shows the desired behavior, with the aid of jit.unpack and jit.pack.

Panel B shows my attempt at simplifying this with jit.expr. I was under the impression that if I apply the @expr to in[0].p[0] I would get the behavior I desire. It appears that jit.expr copies its result to all three output planes rather than just passing in[0].p[1] and in[0].p[2] through.

I'm not understanding something about jit.expr and could use some help.  

In the attached patcher I'm applying a variable scaling via jit.rota with a fixed anchor point.  I want to apply this to plane 0 of a three plane float32 matrix.  

Panel B shows my attempt at simplifying this with jit.expr.  I was under the impression that if I apply the @expr to in[0].p[0] I would get the behavior I desire.  It appears that jit.expr copies its result to all three output planes rather than just passing in[0].p[1] and in[0].p[2] through.

jit-expr-2

it is expected behavior. if you don't supply an expr for each plane, the single expr is used for every plane. the solution would be to provide an expression for plane 1 and 2 in addition to plane 0, which in your case is simply the input matrix.

it is expected behavior. if you don't supply an expr for each plane, the single expr is used for every plane.  the solution would be to provide an expression for plane 1 and 2 in addition to plane 0, which in your case is simply the input matrix.


Thanks Rob, that's it. I didn't realize you had to specify an expr for each plane.

Thanks Rob, that's it.  I didn't realize you had to specify an expr for each plane.
