Re: value-preserving transformations

[Christian Keil <c.keil@xxxxxxxxxxxxx>]

Scott Ferson wrote:
> So, if I understand you correctly,
>
>        [-2,3] * [-2,3] should *not *be optimized to [-2,3]**2
Yes.

> but
>
>        A * A *should *be optimized to A**2.
>
> I think this makes good sense.  In the former case, the compiler has no 
> way to know that the first and second instance of [-2,3] refer to the 
> same quantity, whereas, in the latter case, the programmer has asserted 
> that they do by choosing the one variable name.
I'd replace *should* with *may*. But these transformations should be 
clearly documented somewhere I think.
I think Arnold's proposal is good in requiring that all real, exact 
results have to be enclosed (which would be violated by optimizing 
[-2,3] * [-2,3] to [-2,3]**2) and that the user shall be able to turn 
off optimizations for parts of his program.

Christian

--
                             /"\
Christian Keil               \ /    ASCII Ribbon Campaign
mail:c.keil@xxxxxxxxxxxxx     X  against HTML email & vCards
                             / \