Re: novice: mapcan use?
- From: anamax@xxxxxxxxxxxxx
- Date: 31 Aug 2005 07:56:56 -0700
Peter Seibel wrote:
> Kenny Tilton <ktilton@xxxxxxxxxx> writes:
> > Uh, yeah, actually it does. You are simply defending ignorance. In any
> > given situation I can safely use delete or I cannnot. If I can use
> > delete, then using remove conses for no better reason other than to
> > save me from thinking.
Saving me from thinking is always a good thing.
Structure-sharing is never on the feature list. Certain cases are
performance enablers, but those are the only ones worth thought, at
least for those of us who are overcommitted.
> Well it's possible that using a destructive (or, as I prefer,
> "recycling") function will slow things down--by recycling old conses
> you may move them out of the youngest generation in a generational
> garbage collector where they could have been collected much more
> efficiently than they will now. Or so I understand but IANAGCI.
A copy can be more cache-friendly, which is often more important.
On reasonable-performance systems, floating-point cosine is much
cheaper than an L2 cache miss, so if you're optimizing for operation
count, you're not optimizing for performance. It's the software analog
to "it's not the gates, it's the wires".
.
- References:
- novice: mapcan use?
- From: Bernd Schmitt
- Re: novice: mapcan use?
- From: Pascal Costanza
- Re: novice: mapcan use?
- From: Bernd Schmitt
- Re: novice: mapcan use?
- From: Pascal Costanza
- Re: novice: mapcan use?
- From: Kenny Tilton
- Re: novice: mapcan use?
- From: Pascal Costanza
- Re: novice: mapcan use?
- From: Kenny Tilton
- Re: novice: mapcan use?
- From: Peter Seibel
- novice: mapcan use?
- Prev by Date: Re: Scams
- Previous by thread: Re: novice: mapcan use?
- Next by thread: Re: novice: mapcan use?
- Index(es):