Big TP codespace

[Martin Duke <[e-mail]>]

Here's an annoying part of the transport parameter spec:

There are 2^16 possible TPs and I should handle them all gracefully, but
the highest standard one is 13. I also MUST blow up the connection if
there's a duplicate TP. For 13 TPs, I can very easily check for duplicates
with a 16-bit field. But to meet this requirement, I need an 8KB field or a
fancier, more time-consuming data structure, which is really cumbersome.

If I'm actually supporting a handful of experimental TPs, it's easy to
track those by exception. What I'd propose is that we MUST end the
connection if there's a duplicate of any of the TPs I actually support.

What do people think?