If they're dealing with non-native architectures, they might have bigger problems than endianness. Unpack will only be useful if the both platforms use the same format for floats.

I assumed they are on an x86 or x86-64, as I would have expected them to mention it if it wasn't the case.

I also assumed that they were referring to IEEE double-precision floats. Though I did give them the information they needed for IEEE single-precision floats too via the link.