This is the mail archive of the guile@cygnus.com mailing list for the guile project.
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Maciej Stachowiak writes: > > forcer@mindless.com writes: > > This happens on a 32bit box: > > > > guile> (logand #b1000000000000000000000000000000 0) > > 0 > > guile> (logand #b10000000000000000000000000000000 0) > > ERROR: In procedure logand in expression (logand 2147483648 0): > > ERROR: Wrong type argument in position 1: 2147483648 > > ABORT: (wrong-type-arg) > > > > I conclude from this this that Guile is unable to handle 32+ bit numbers > > with logical functions, is this correct? > > Since i need this (i'm trying to implement CAST-128 encryption as per RFC2144) > > i wonder wether anyone found any solution to this problem already > > short of using vectors for the bit patterns? > > I know i could patch the guile sources to include a CAST encryption function, > > but i'd prefer a Scheme-only solution. > > Any help is appreciated, and thanks in advance, > > Guile can logically and anything up to the range of a signed long on > the target platform. IMO, it should use unsigned long instead as a > stopgap; ultimately, the logical boolean functions should be extended > to handle bignums, again IMO. I suspect Jim would accept a patch to > that effect. I would really like such a feature myself (scwm goes > through some gyrations to do bit-twiddling on 32-bit unsigned > quantities) and would code a patch if I had time. Ouch that breaks the bitflag-based sysinterned constants in my wrapped curses modules (highlighting attributes etc. are encoded as 32-bitmasks in ncurses). Klaus Schilling