Re: [PATCH v2 0/6] MR2722: dsound: Doppler shift (part 2)

[Anton Baskanov (＠baskanov)]

On Mon May 22 06:57:13 2023 +0000, Huw Davies wrote:
> I was expecting running the new test before the final commit to crash
> with a divide by zero which we should avoid.  I actually don't observe
> this (even with `WINETEST_INTERACTIVE` set) - any idea why that would be?

It's a floating point division, so instead of crashing it produces a NaN. When converted to DWORD, NaN becomes 0. Though, according to C99, the value is unspecified, and with really big values we can crash the mixer.