From: Dan Sugalski <dan@...>
Date: 2003-01-06T11:15:46+09:00
Subject: Re: Compile time constant folding?

At 9:32 AM +0900 1/6/03, Yukihiro Matsumoto wrote:
>Hi,
>
>In message "Compile time constant folding?"
>     on 03/01/06, Dan Sugalski <dan@sidhe.org> writes:
>
>|Here's a quick question, now that parrot's close to getting object
>|support and the cardinal folks are on their way to a working parser...
>|
>|Is it valid to do compile-time constant folding as an optimization?
>
>I didn't take it, because
>
>   * every method may be redefined (e.g. "+") in dynamic languages.
>
>   * besides, foldable expressions are rather simple, so that it does
>     not save much time.

You'd be surprised, especially when you factor in some other 
optimizations--you can do some substitutions and inferences of 
variable contents. Because of it, some expressions that look like 
they're variable, such as:

	a = 12
	b = 8
	c = a + b

are really constant. (And yes, that is really cut down and 
excessively simplified)

Still, if it's not allowed that's fine. The question came up and, 
since I have a bunch of other things I should do first, I couldn't 
let it go without finding out.

Thanks, Matz.

>The latter assumption may not be true for Parrot, which has much
>faster and more optimized runtime.

We'll see--you still need to beat us with sprite. ;)
-- 
                                         Dan

--------------------------------------"it's like this"-------------------
Dan Sugalski                          even samurai
dan@sidhe.org                         have teddy bears and even
                                       teddy bears get drunk