Saturday, May 29, 2004
Concurrent Haskell
http://citeseer.ist.psu.edu/jones96concurrent.html
full paper at
http://research.microsoft.com/Users/simonpj/Papers/papers.html
This paper needs more examples.
How do you:
write a fork bomb?
do symmetric forks (symFork)?
do lazy-list communications?
non-deterministic join?
non-deterministic split?
wait for the termination of a process?
kill another process?
what are the types of getChan and putChan?
what is unGetChan supposed to do?
Where's the "full version" of the paper as referred to in
Section 4. There it is:
http://research.microsoft.com/Users/simonpj/Papers/papers.html
In any instance of takeMVar foo >> putMVar foo (e.g., figure
2), what if the actions aren't atomic?
example of priority?
I really don't "get" this "singular choice" thing.
Friday, May 28, 2004
128 bit key exchange
? nn=2^128-159 %89 = 340282366920938463463374607431768211297 ? factor(nn) %90 = [340282366920938463463374607431768211297 1] ? znprimroot(nn) %91 = Mod(5, 340282366920938463463374607431768211297)So 5 can be the base for diffie-hellman key exchange. Missing 159 or so keys, but that's such a small fraction of the keyspace, so who cares.
? nn=2^256-189 %92 = 115792089237316195423570985008687907853269984665640564039457584007913129639747 ? factor(nn) %93 = [115792089237316195423570985008687907853269984665640564039457584007913129639747 1] ? znprimroot(nn) %94 = Mod(2, 115792089237316195423570985008687907853269984665640564039457584007913129639747)Similarly 2 can be the base for 256 bit keys. ---- Too bad such small primes can be broken with current methods of finding discrete logs. ---- Elliptic Curve Diffie Hellman would work, though.
Tuesday, May 25, 2004
Friday, May 21, 2004
More morse code
ABCDEFGHIJLKMNOPQRSTUVWXYZ = 26
lowercase = 26
int'l characters = 8
lowercase = 8
numbers = 10
punctuation = 11
space = 1
other-paren = 1
91
and still wanting newline and end-of-file
93
removing lowercase
59
the ui elegance of twoinputs gets two outputs (16*16=256).
slider and 10 buttons?
12345 67890
QWERT YUIOP
asdfg hjkl [spc]
zxcvb nm,./ (3 punc, so 7 remain not counting bracket)
88888 888 PP
PPPPP () [newline] [eom]
magic squares
| 0 | 7 | 9 | 14 |
| 11 | 12 | 2 | 5 |
| 6 | 1 | 15 | 8 |
| 13 | 10 | 4 | 3 |
Unclaimed Money and Assets: Questions and Answers: Financial Management Service
Unclaimed Money and Assets: Questions and Answers: Financial Management Service
The feds give some slightly useful information.
Who is this www.onlinecashrecovery.com a.k.a., "U.S. Claims Service" claiming to have 128 dollars for me and why don't they have any links from google?
Google Search: link:www.onlinecashrecovery.com
Google Search: onlinecashrecovery
whois tells me
-----
Registrant:
U.S. Claim Services LLC
3612 Coffee Rd. Suite B
Bakersfield, CA 93308
US
Domain name: ONLINECASHRECOVERY.COM
Administrative Contact:
Admin, Domain domains@themarcomgroup.com
3612 Coffee Rd. Suite B
Bakersfield, CA 93308
US
661-589-2076
Technical Contact:
Admin, Domain domains@themarcomgroup.com
3612 Coffee Rd. Suite B
Bakersfield, CA 93308
US
661-589-2076
Registration Service Provider:
The Marcom Group, Incorporated, domains@themarcomgroup.com
661-589-2076
This company may be contacted for domain login/passwords,
DNS/Nameserver changes, and general domain support questions.
Registrar of Record: TUCOWS, INC.
Record last updated on 29-Apr-2004.
Record expires on 29-Apr-2005.
Record created on 29-Apr-2004.
-----
so it's new. (28 Apr 2004). It seems
to be run out of the same address as the web design firm which
is a little weird, or maybe it's a convenience/laziness of the
domain registration.
I wonder how they got so many true stories testimonials if they've only existed for 20 days so far.
They say "Private Investigators License PI 12555 and our Business License 0458907". Oh google searching for "0458907" is useful, returning www.usclaimsservices.com which whois tells me:
% whois usclaimsservices.com
[Querying whois.internic.net]
[Redirected to whois.melbourneit.com]
[Querying whois.melbourneit.com]
[whois.melbourneit.com]
Domain Name.......... usclaimsservices.com
Creation Date........ 2003-10-02
Registration Date.... 2003-10-02
Expiry Date.......... 2005-10-02
Organisation Name.... us claims services
Organisation Address. 2901 f street
Organisation Address.
Organisation Address. bakersfield
Organisation Address. 93301
Organisation Address. ca
Organisation Address. UNITED STATES
Admin Name........... Paul Hashim
Admin Address........ 2901 f street
Admin Address........
Admin Address........ bakersfield
Admin Address........ 93301
Admin Address........ ca
Admin Address........ UNITED STATES
Admin Email.......... phashim@stoppreapproved.com
Admin Phone.......... +661.6332416
Admin Fax............ +661.6332457
Tech Name............ Paul Hashim
Tech Address......... 2901 f street
Tech Address.........
Tech Address......... bakersfield
Tech Address......... 93301
Tech Address......... ca
Tech Address......... UNITED STATES
Tech Email........... phashim@stoppreapproved.com
Tech Phone........... +661.6332416
Tech Fax............. +661.6332457
Name Server.......... ns1.themarcomgroup.com
Name Server.......... ns2.themarcomgroup.com
October 2003 (7 months ago) somewhat more reasonable.
And now some link backs to
2901 f street and 2901 f st. Something about chiropractors.
"Paul Hashim" has a link to the sale of retail property at 2901 f street.
The Plano TX connection also appears here:
search on openrbl for onlinecashrecovery.com
Block Cipher to Stream cipher
Block ciphers can be converted into stream ciphers by xoring
with the encrypted plaintext like 1,2,3,4,5,etc. so we
don't need to use RC4 and all it's FUD about weak keys and
weak initial 256 bits and even the fact that it's trapped in
binary-land.
Though it does ask an interesting mathematical question:
Given a stream of bits, how can we convert it losslessly
(non-wastefully) into another base?
Wastefully, for example, if we wanted to convert to base-65,
we can read 7 bits at a time and throw away values from
65..127. But non-wastefully?
Thursday, May 20, 2004
or just screw it all, a stream cipher is better, like
enigma was, so where's my RC4 decoder ring?
56 buttons of input, plus END-of-message and a "cipher me" button
the wasted extra bits can be used to store entropy.
27 lights of output (27th is a button indicating "more",
so keep hitting "cipher me" until the more
light turns off)
or better, two rows of 16 (maybe the sixteen simple morse)
16*16=256. so one char of input= 2 chars of output.
a bunch (39?) of cylinder mounted dials indicating the key
at the top. it automatically resets itself to zero
when "load key" happens.
. 1 E
.. 1 I
... _ 2 S T
.... ._ _. 3 H A N
..... .._ ._. _.. 4 5 U R D
...... (nonexistent)
_... B
._.. L
.._. F
..._ V
__ M
6
17 morse codes total with length less than or equal to 6 ( .=1, _=3 )
>> 128*log(2)/log(17) = 31.3153
or using the sixteen short existing morse codes "eist han5 urdb lfvm"
(a little unfortunate that s and 5 are both there)
(and u v) (and i l)
log16 = 32
My AES decoder ring
input = 55 international morse code + space = 56 chars
22 can fit in a 128bit bit block
(allows up to 56, to room to spare)
base 68 chars fits 21 in a 128bit block.
output and key = only use 26 chars
= 28 (27.2) to transmit a 128bit block
or
>> 128*log(2)/log(10) = 38.5318
39 7-segment displays?
>> 128*log(2)/log(27) = 26.9197
>> 128*log(2)/log(37) = 24.5707
can someone build a physical device which might
look like a cereal box decoder ring or a
german Enigma? I'm imagining 22 sliders
and maybe N 1-char output displays. N=39 is promising.
"It's like an Enigma, except it uses Rijndael."
"how to implement rijndael mechanically"
Of course one should be doing ciphertext chaining
and initialization vectors, etc, etc.
maybe one setting of of the input sliders generates a random number.
Subscribe to:
Posts
(
Atom
)