The question has to be; is what being fixed - the initial discrepancies (between platforms, size of floats, and with respect to documentation), or 'nan' propagation ?Having nan propagate correctly means I don't have to check at every step like I do when using -99999 .
Is this being fixed in the SDK ? It's a shame to lose that extra speed and efficiency when it's needed.
Most time it needs a bug report or issue raising to get something fixed. I am not sure anyone has previously noticed or noted that 'nan' doesn't propagate correctly.
Statistics: Posted by hippy — Mon Oct 27, 2025 1:21 am