Re: [9fans] Floating point and dividing by 0

[Riddler <riddler876@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1076 bytes --]

I imagine I was expecting a bit much. I was expecting that it would execute
the instruction, thus putting infinity in the variable, then trigger a
"divide by 0" note. Meaning if I continued infinity would be there.

Thinking about it in hindsight, I could see the CPU/FPU/X doing "get
instruction -> check divisor -> trigger interrupt". With fcr deciding if
the last step is "trigger interrupt" (assuming interrupts are triggering
the note) or "return infinity".
​
I can certainly see potential for lots of hardware edge cases, and why
doing this the way I am probably isn't a safe.

Thinking about the 'skipping an instruction' thing. I imagine you have to
or you would never be able to continue from any note that was triggered due
to an instruction in the program. At least not without some logic in the
note system to check what triggered it, where and ignore it if needed.
Wouldn't be so keen on it if the note was triggered by another program or
something though.
Might go look at the note system, I'm curious how these situations are
handled now.

[-- Attachment #2: Type: text/html, Size: 1231 bytes --]