TSRPv1 spec draft

[olorin]

The spec itself is just a draft and quite possibly has errors, but I'm reasonably happy with the protocol as a whole now. Note that there have been two changes to the protocol since last week:

• Removed the ability to specify the hash function. (#24, #25)
• Removed the prefix string "TSRPv1-begin" from the auth key derivation. We just want to mix the protocol identifier into the auth key bitstring, there's no reason to do it twice. (#27)

@markhibberd @erikd-ambiata @thumphries @charleso

Fixes #13.

[olorin]

Another question: what format do we want to give this to customers in? The current version is markdown-flavoured plaintext; I'm good with that but I don't know if we're going to want to put this on an HTML API docs page or the like. Probably not worth worrying about for now but just thought I'd mention it in case people have opinions.

[charleso]

@olorin Would love an in-person walk through of this when you're in the office next (just to improve my chances of absorbing it).

[olorin]

Would love an in-person walk through of this when you're in the office next (just to improve my chances of absorbing it).

Excellent idea, that will be good to make sure what I put in the spec is the same as what I have in my head as well. I'll be in today late morning/early afternoon (whenever my partner wakes up, she's been giving me a lift in on the way to Google).

[erikd-ambiata]

:ice_cream:

[olorin]

Everyone good with this after the IRL overview? @markhibberd @charleso @thumphries @nhibberd

[thumphries]

Yep 0️⃣

[charleso]

👍