Divide operator

John Abshier

I unintentionally divided an integer by zero in Spin and learned the following: x/0 returns 0.· I am undecided if this is a bug or feature.· Other options would be to·abort (If you don't guard for division by zero you shall be punished.) or POSX, NEGX depending on the sign of x.

John Abshier

Mike Green

There is no "standard" result for x/0. Mathematically it's undefined. Practically speaking, none of the possibilities you suggested (abort, POSX, NEGX) really provide a reliable answer. There's no interrupt for divide by zero as exists in some microprocessors. Returning POSX or NEGX wouldn't provide any more meaningful answer if the program doesn't check for these. It's not like these are special propagating values as is found in some floating point implementations (NaN). You really have to check for a zero divisor or design your program so a zero divisor can't occur.

VIRAND

And there's nothing to abort to.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I should be typing in Spin now.
Coming soon. My open Propeller Project Pages and favorite links index.

mpark

Doesn't an abort with no "catch" just halt the Spin interpreter?

John Abshier

Mike, I agree that checking and design are the way to go. It caught me a little by surprise because my background is with languages that produce an error on zero divide since it is mathematically undefined. Just brought back painful memories of long ago when my program would not work because +0 was not equal to -0.

John Abshier

Dave Hein

I think the 32 least significant bits of infinity are zero, so this result is correct.

It might be nice to provide for a user defined exception handler.· This would be useful for floating point exceptions and asserts.