[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: End to end mixminion rationalized, with compression and K-of-N
On Tue, 2002-10-15 at 04:39, Roger Dingledine wrote:
[...]
> Since we already say 'tag' in the context of a tagging attack, can we use
> an alternate name for this? id? mid (mixminion id)? something better? :)
[...]
> State-keeping
> surb? State-carrying surb? We may be able to decide a better name here
> once we've renamed 'tag' above.
I have no attachment to the current names. I'll try to think of some
better ones; other people should think too, since I might not come up
with anything.
> > Let M = M | PRNG(Len(M) - NUM_CHUNKS*PS*K)
>
> Is the arg to PRNG going to be negative here?
Oops. You're right. The operands of the subtraction should be
switched.
> > For i from 0 to NUM_CHUNKS - 1:
> > Let CHUNK_i = M[i*PS*K : (i+1)*PS*K]
>
> If we're using the same notation as before, this should be M[i*PS*K :
> PS*K], yes? I think I argued a while ago we might want to change to
> your above notation. Should we?
Ick; I misremembered and thought we had. Honestly, I like my current
notation better, but we can do better than argue about notations. Is
there a major precedent for either one? (I'm modeling my current one
after slicing in Python, Matlab, etc.)
[...]
> > Note that a message will rely intact if and only if at least K
>
> 'rely'? You mean 'arrive'? 'remain'?
"Arrive."
[...]
> > THE DESIGN ITSELF
>
> Bedtime for me. I'll read the meat of this tomorrow. :)
Cool.
By the way, I found an error in my spec when I was implementing it. In
section "C" the following text is wrong:
If message-type 'PLAIN' or 'REPLY', let SZ = 28K. Else let
SZ= 28K-42bytes. (SZ is the amount of data that fits
in a single payload. 'ENCRYPT' messages lose 42 bytes
to OAEP padding.)
It should say instead:
If message-type 'PLAIN' or 'REPLY', let SZ = 28K. Else let
SZ= 28K-38bytes. (SZ is the amount of data that fits
in a single payload. 'ENCRYPT' message lose 42 bytes to
OAEP padding, gain 20 bytes by stretching the RSA-encrypted
data into the Tag field, and lose 16 bytes by storing the
session key.)
--
Nick