Re: [clisp-list] Large performance decrease observed with timingloop

[Adam W. <li...@co...>]

On Mon, 2002-06-10 at 13:22, Pierpaolo BERNARDI wrote:
> From: "Adam Warner" <li...@co...>
> 
> > Thank you for your contribution which does scale linearly. BTW it
> > generates the wrong result:
> 
> Just add a NREVERSE:
> 
>               (apply #'concatenate 'string (nreverse l)))
> 
> > But that's not the point. The point is to understand why performance
> > decreases so markedly with my version and whether this is unavoidable.
> 
> I showed you how to avoid it.  

Well you showed me how to create superior code. I just though that my
inefficient code was causing a compiler slowdown. I wasn't looking for
superior code. I thought I was highlighting an issue with the clisp
compiler. Turned out I was wrong.

But thank you Bernardi and Don. I managed to get your names right this
time :-)

> Now look closely at your version. Then at mine.  
> Notice the differences.

That's fine. The problem was I suspected those differences were critical
in understanding what was causing the compiler slowdown. But it turned
out my code just kept generating larger lists instead of the same list
each time.

Regards,
Adam