Re: Interop lsh and SSH-2.0-GitLab-SSHD

> You're not gaining anything by doing this apart from (a) wasting a
> ton of CPU cycles [*]

> [*] This may be why your SSH handshake is so surprisingly slow.

Is it?  I'd have to go reread the protocol, but I think at least the
authentication part is doing a _private_ key operation, whose exponent
will be roughly the same size as n - or, possibly, two operations with
exponent sizes comparable to p and q; it's been long enough I no longer
remember which operations can take advantage of the CRT.

> and (b) risking non-interoperability with other implementations that
> enforce sanity-checks on public-key values.

As far as I'm concerned, artificial refusal to interoperate with a peer
that's conforming to the protocol is the fault of the implementation
imposing the artifical limitation.  (Not that that's necessarily a bad
thing; my world-facing ssh servers do it under some circumstances.  But
I don't blame the peer for those interoperatbility failures, not at the
ssh level.)

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B