>>25
I wouldn't say measuring it in bytes is wrong. To meaningfully deserialize it on a machine with different byte size, the encoded byte size would still have to be known. So it makes sense you would need either a (file size in bytes, byte size in bits) or (file size in bits, byte size in bits) pair. Either can be derived from the other two.
It still "isn't nessecarily wrong to say a file is x bytes either, as long as you know the number of bits in a byte on the architecture the file was created." The point I was trying to make is, it currently makes no difference. For all intents and purposes, right now a byte
is an octet.
I appreciate where you're coming from, trying to put it into it's simplest and most intuitive form. I'm arguing the opposite to show you that it isn't wrong either.
This is remarkably similar to the Pi/Tau debate.