Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Pierpaolo Bernardi
On Sat, Nov 24, 2012 at 8:33 PM, Jens Axel Søgaard wrote: > Hi All, > > I have written an implementation of bit vectors intended to be part of > the data collection. > > https://github.com/plt/racket/pull/176 > > Any comments on the implementation and documentation are welcome. > The bit vecto

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Jens Axel Søgaard
Hi Pierpaolo, 2012/11/27 Pierpaolo Bernardi : >> Any comments on the implementation and documentation are welcome. >> The bit vector is represented as a vector of fixnums (packaged in a >> struct of course). > > I seem to understand that you do not exploit the packed representation > of bits in th

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Pierpaolo Bernardi
Hi, On Tue, Nov 27, 2012 at 2:34 PM, Jens Axel Søgaard wrote: > Hi Pierpaolo, > > 2012/11/27 Pierpaolo Bernardi : >> Also, you store and retrieve booleans, not bits, so the name >> 'bit-vector' is misleading. > > Potato / Potato :-) > > http://wiki.call-cc.org/eggref/4/iset Excellent! so now I

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Matthew Flatt
Nicely done. I've merged with minor changes, including renaming `bit-vector-count' to `bit-vector-length' to be more consistent with `vector' functions. At Sat, 24 Nov 2012 20:33:12 +0100, Jens Axel Søgaard wrote: > Hi All, > > I have written an implementation of bit vectors intended to be part o

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread J. Ian Johnson
As I intend to use bitvectors to do fast set operations, and cardinality is a set operation, I wrote up a "fast count bits" function that should be rolled in for the vector implementation: https://gist.github.com/4154642 It depends on the fixnum representation, so I don't imagine this being ada

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Jens Axel Søgaard
2012/11/27 J. Ian Johnson : > As I intend to use bitvectors to do fast set operations, and cardinality is a > set operation, I wrote up a "fast count bits" function that should be rolled > in for the vector implementation: > https://gist.github.com/4154642 I have used your code to add a function

Re: [racket-dev] Implementation of bit vectors

2012-11-27 Thread Sam Tobin-Hochstadt
On Tue, Nov 27, 2012 at 1:15 PM, Jens Axel Søgaard wrote: > > There is an issue of potential confusion over names though. > In the data collection, the -count suffix normally returns the size of > the data structure. > For vectors the suffix -length is normally used. The name 'popcount' is common