Re: [Sbcl-devel] SBCL chokes on large case statement...

[Kevin R. <kp...@ma...>]

On Feb 15, 2009, at 16:41, Paul Khuong wrote:

> It is a design issue with the compiler and the way it uses static
> analyses. Neither safety nor your observations on radically different
> implementations have anything to do with it. CMUCL and SBCL are known
> for their compiler's sophistication. Sophisticated analyses
> unfortunately tend to come with additional computational cost at
> compile-time.

How about explicit upper bounds in the compiler as to how much  
thinking about the code it will do? (Limits in terms of explicitly  
defined steps, not CPU time, so as to be reasonably deterministic.)

The limits could be twiddled by declarations, COMPILATION-SPEED or  
SBCL-specific ones.

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>