Hi Gil,

the reason, why it is wise to place the stack on the end, was, that its real length was unknown and difficult to calculate (depending from the software complexity)
But all data/idata variables always have a defined length, e.g.:

int idata val[4]; // consume 8 bytes of idata

To reverse this order cause always, there is unused memory on the end. And if the stack grows more than expected, it corrupts data instead using this unused space.
Also corrupting data was harder to detect, than a complete crash.
And mostly, if the stack exceed the physical memory limit, it crashes completely on the attempt to return.


Peter