Re: judy with other data structures

[Alan S. <aj...@fr...>]

Richard,

> I am developing some software (a sparse volumetric array) that
> requires an associative array with integer key, value pairs.

Sounds like a fine use of libJudy.

> I am considering using the Judy array but the software I'm developing
> uses up a fair amount of memory >256MB, of which the associative array
> would only be a small part (~20MB).  My question is will the memory
> usage by the other data structures impact negatively on Judy?

Shouldn't make any difference -- why would it?

> I thought the Judy array was fast as it was very compact, but will
> that be negated by the other stuff.  Or do I not really know what I'm
> on about.

Well, insofar as you store anything in a libJudy array, it will be
relatively efficient, if you measure bytes-per-index.  Underlying data
to which it references/points, etc, that's your problem.

It should be pretty easy to try it and see, once you get past the small
learning curve on how to call the libJudy functions or macros.

As one of the creators of libJudy in 2000-2002, I'm enjoying presently,
at my new job, actually making regular use of it, and very effectively.
Need a quick crossref/quickref?  Easy to create, use, destroy; few lines
of code, and seldom any bugs in my calling sequence (as long as I get
the pointer right).

That's in straight C...  It might actually be easier -- although
significantly less efficient?  -- in other languages like C++.

Cheers,
Alan Silverstein