Right now we're doing some evil things with bitmasks on pointers to find
the base address of the register chunk structure given a pointer to any
place in the structure. I'm about to do this for stack chunks as well.
Unfortunately, what we're doing now (with a hard-coded constant in
config_h.in) is wrong, as it makes some serious assumptions about the
constancy of size of the register chunk. It can't do that, since the size
will change depending on the size of ints, floats, and pointers on each
individual platform.
Anyone care to throw together some tests for configure.pl to figure out the
*real* sizes for the register chunks (courtesy of a quick test program that
does some sizeof() checks?) and figure out the power-of-two boundary these
things really live on?
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
[EMAIL PROTECTED] have teddy bears and even
teddy bears get drunk
