From: Nicolas C. <war...@fr...> - 2004-03-20 10:03:46
|
> I made a few changes on the files bitSet.ml and bitSet.mli (see > attachements). I : > > 1 - corrected a few orthograph mistakes (I hope I didn't added new ones) > 2 - corrected an "lsl" -> "lsr" (the size of a BitSet was 8 times bigger > instead of 8 times smaller) > 3 - suggested the name "copy" which is more usual than "clone" > 4 - used "blit_string" and "fill_string" for more efficiency > 5 - added some functions on sets > > I hope there are no more mistakes. I have got a question : isn't it > possible to use the 32 bits of an int (or 64 bits, depending on the > architecture) for efficients BitSets ? The String functions seem to work > on 8 bits only. Thanks, The CVS have been updated with your changes. To answer more precisely to your questions : - we could actualy use ints as 16 bits values if we add the API to extract them from strings, but we don't :) - we can't use Pervasives.compare because bitsets might be of different sizes but are actually the same integer - I disabled the "deprecated" on "clone", having both copy and clone is okay. Regards, Nicolas Cannasse |