On Wed, Jun 27, 2012 at 3:16 PM, grarpamp <[email protected]> wrote:
> What is use case of 95 bit choice?

Unique local IPv6 prefix is 7 bit, and 95 bit for the hash leaves 26
extra bits in IPv6 to avoid issues with any other mapping — looks
about right to me. Also, base-32 addresses are optimal with
multiple-of-5 bit counts.

> Short hashes unavailable, so it would be truncation, which made
> from a secure hash probably better than junk short hash anyway.

Sure, I had some truncation method in mind, or even plain truncation
like Tor does (but — didn't see any second-preimage attack resistance
analysis of plain SHA-1 80-bit truncation in Tor specs).

> I had also thought I2P could generate keys, but reject putting
> into operation those that fall outside of some desired mask
> range, say only use first 80 bit of 256 space.

That's computationally infeasible (essentially what Bitcoin relies on).

-- 
Maxim Kammerer
Liberté Linux: http://dee.su/liberte
_______________________________________________
tor-talk mailing list
[email protected]
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk

Reply via email to