Ignoring Alpha OpenVMS arithmetic error

[Deleted_U_Intel]

Is there any way on such a system to ignore certain arithmetic errors through environment settings rather than by compilation options?

I am getting a high performance arithmetic error for what it detects as a divide by zero. (I know, sloppy code). The same code on Windows or various Unix systems works -- it apparently returns a zero, which is OK.

The issue is that this was compiled on a customer system on the other side of the world and we do not have an Alpha system or essentially any OpenVMS knowledge.

Thanks for any help.

[Steven_L_Intel1]

I suspect that on the other systems, you get an IEEE NaN, not a zero. You can get that too on VMS, but need to compile with /FLOAT_TYPE=IEEE /IEEE_MODE=UNDERFLOW_TO_ZERO There is no environment setting that will do this for you, as it requires completely different generated code.

Steve