We have a legacy Cocoa library of mathematical algorithms that has worked
fine since OS X 10.6, but running this same code on 10.10 results in odd
numerical errors (that is, incorrect results). I'm thinking this could be
the result of differences in ILP32 vs LP64?
The code variables are ints and pointers, and there's a bunch of memory
allocation for arrays that looks fragile. I know that the size of ints has
not changed, but pointers have gone from 4 bytes to 8 bytes in LP64. Could
this be the problem?
Here's a sample of affected code:
int ***hov;
/* memory allocation for data arrays */
if((hov=(int ***)malloc(LNUM*sizeof(int **)))==NULL)
{
fprintf(stderr,"Error in memory allocation for hov!\n");
return(-1);
}
for(k=0;k<LNUM;k++)
{
if((hov[k]=(int **)malloc(XDIM*sizeof(int *)))==NULL)
{
fprintf(stderr,"Error in memory allocation for hov[%d]!\n",k);
return(-1);
}
for(i=0;i<XDIM;i++)
{
if((hov[k][i]=(int *)malloc(YDIM*sizeof(int)))==NULL)
{
fprintf(stderr," in memory allocation for hov[%d][%d]!\n",k,i);
return(-1);
}
}
}
/* calculate total field in deg^2 tested per level */
TOTAL_VF=XDIM*YDIM*(GRES/60.0)*(GRES/60.0);
. . . etc.
I've tried promoting the ints to longs, but the errors in the mathematical
computations became much worse.
-Carl
_______________________________________________
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
