[Mono-list] [Mono-dev] faster than hashtable?

[Felipe Lessa]

On Wed, Oct 28, 2009 at 03:10:41PM -0200, Rafael Teixeira wrote:
> As p/invokes cost a lot, another (fourth) option would be to
> reimplement judy arrays algorithm (which I admittedly know nearly
> nothing about) in managed code.

I *really* wouldn't think of that before being damn sure that
those calls are a performance problem :).  There are dozens of
different data structures for each word size (32 or 64 bits),
many many rules between them and, of course, optimizations.  It
may be an interesting exercise, but nothing more than that.

Actually, I hear some people dislike Judy because of its
complexity.  They prefer to use simpler solutions even if Judy's
API is as simple as any other data structure's.

--
Felipe.