dmd foreach loops throw exceptions on invalid UTF sequences, use replacementDchar instead

FeepingCreature feepingcreature at gmail.com
Fri Nov 5 12:13:17 UTC 2021


On Friday, 5 November 2021 at 12:03:24 UTC, Ola Fosheim Grøstad 
wrote:
> For instance in raytracing I would want NaNs. Then I can make a 
> choice based on neighbouring pixels whether I want to compute 
> it again using a slower method or simply fill it in with the 
> average of the neighbours (if all the neighbours have roughly 
> the same colour).

I can't imagine wanting nans in raytracing. Just the idea of a 
fpu slowpath-provoking nan making its way into my nice wide SSE 
vectors gives me hives. Any sensible raytracing routine should 
just never produce a nan to begin with.

(For denormals there's FTZ/DAZ, at least.)


More information about the Digitalmars-d mailing list