8-bit clean

✎ This article/section is a stub — some half-sorted notes, not necessarily checked, not necessarily correct. Feel free to ignore, or tell me about it.

If the answer to "can we pass 8 bits in a byte safely", it is 8-bit safe, a.k.a. 8-bit clean.

Historically there were serial/modem lines that sent 7 bits, e.g. where people assumed the data never used it (e.g. is pure ASCII), which was a little cheaper where networking was expensive (but this argument actually became somewhat irrelevant relatively early in the modem era).

A number of early internet protocols (e.g. mail and news) required the eighth bit not be used, either never or not until requested.

If you start counting protocols where some parts are ASCII only then the list gets longer, e.g. includes FTP in text mode (though that's rare to use), parts of DNS, HTTP headers.

Perhaps the largest example today is email.

Even though these days all parts (storage, TCP) are 8-bit clean, client and server must still agree on this.

In practice 8-bit is almost always negotiated(verify). Technically, clients you should still understand its absence. And yes, if not negotiated, the client should encode the body differently (and less efficiently)., Note that this also applies to each MTA-to-MTA send, though it's rare to see them downgrade, and some MTAs don't even do that re-encoding).