PING/PONG ambiguity

[mkieweg]

The definition of PING and PONG frames are ambiguous.

Section 8.9 states:

If the Data field is not empty, the recipient of this frame MUST generate a
PONG frame (Section 8.15) containing the same Data.

Section 8.15 states:

If the content of a PONG frame does not match the content of a PING frame
previously sent by the endpoint, the endpoint MAY generate a connection error of type
UNSOLICITED_PONG.

The way I see it either the MAY in Section 8.15 has to be a MUST to solve this ambiguity or the part in Section 8.9 has to be changed into something less strict.
This would be my preferred option since it gives the PING/PONG frames a little bit more semantic leeway.

[mikkelfj]

Even if the PONG'er behaves correctly, arbitrary network delays may cause PONG packets to arrive after they are no longer remembered by the PING'er. Requiring a MUST makes correct evalutation much harder, either by storing PING frames indefinitely or by using some hashing to confirm the validity of the PONG frame.

[martinthomson]

@mikkelfj's answer here is the reason that this isn't a MUST. @boisjacques, do you think that there needs to be some better explanation of this?

[mkieweg]

Thank you for the clarification, for me it's clear now. However it might help to make the implicit explanation more explicit.

[ianswett]

PING and PONG have been replaced by PING, PATH_CHALLENGE, and PATH_RESPONSE, so hopefully that closes this issue?

[MikeBishop]

The text is still there, but I agree with @mikkelfj and @martinthomson that the text is correct. The sender MUST do the right thing, but the receiver isn't required to track sufficient information to detect it 100% of the time.