Open peterychang opened 5 years ago
Floats are ieee 754, which works all over the place and are well specified.
Thats not guaranteed
float - single precision floating point type. Usually IEEE-754 32 bit floating point type double - double precision floating point type. Usually IEEE-754 64 bit floating point type
They're usually IEEE754, but not guaranteed to be. This won't be a problem for most systems, but I know some embedded systems for example, do not use that standard
Can we define the stored model file as IEEE-754, and any system on which we do not read correctly due to using the wrong float type is buggy?
(Alternatively, converging on an IDL-based serializer would also work here)
@lokitoth @peterychang What do you think is the progress on this issue?
Model files contains bit patterns for certain types of data (like floats). Since the size and the precise bit-wise format of a float is not guaranteed by standards, the resulting model files are technically not portable