bonjour monsieur quessy,

my theory: it runs under i386 under max with doubles internally, the problem is the method binding to PD and it's not a problem with the rest of the struct members. having them as doubles shouldn't prevent compiling AND the results are different if run as 32bit floats or 64bit doubles.

i say try to change just the function definitions: for example void Flock_mode(t_boids *x, double arg); to: void Flock_mode(t_boids *x, t_float arg);

i just did so myself and it compiles and runs fine on Mac OS X/PD extended 0.39-4

i'm attaching the two source files for you to try.

there will be a new version of the boids in Max shortly adding a few features: when that's ready i'll update the PD version and submit it to CVS.

cheers

/*j

Thomas Ouellet Fredericks has found a bug in the [boids2d] external for pd which makes its compilation under GNU/Linux i386 impossible. To fix it, one only needs to replace all the "double" by "float". It might be the same thing for the [boids3d] external. Can I submit this major bugfix in the CVS or will you do it ?

Thanks ! Very nice contribution, by the way.

well, it's not my work exclusively. there have been a few people involved, namely eric singer and andre sier. the original max-object goes way back, it must be at least 10 years old. what andre and i did was adapt it to 3D and make it run in floating point space. my main interest has been to use this as a spatialisation algorithm for multichannel audio: it's very effective.