unforum

[Hobbess]

Thanks for the help addybobble, in the same vein as sparhawk1974 I don't really know what's going on but am a quick learner so keen to see if I can work this out.

Struggling with the pycypto also. Have unpacked the file so can now see the setup.py file. now when I use the next line to build the file get the error:

bash: python: command not found

What should I be doing here? Looking forward to getting it working, even if all I do is work on the practice ones!!

As a further note. when I downloaded the pycrypto file it was called pycrypto-2.0.1.tar.tar (not .gz) could only then run the "tar" line not "unzip"

Edited 11th
Found out i hadn't installed python on the cygwin install, this is now on and getting a problem during the build. Will wait and see what happens with rccrypt etc below.

[addybobble]

Maybe you haven't loaded up Cygwin +before+ you unpacked pycrypto. Sorry, but I'm not going to do IT support for this stuff

Alternatively you must run from the +cygwin+ prompt, not a m$ cmd or command prompt to make this all work.

I'm assuming you are using windows XP or the like.

Incidentally, if you are using a grown up OS like an apple MAC you're out on your own, unfortunately (though no doubt everything you need will be there). It wouldn't be suprising if all the cygwin commands work seamlessly (in fact i think they will): Cygwin adds a rich set of UNIX/LINUX commands to windows, which makes it usable for developers such as myself that use open source code.

Also, one other thing (last help for the day), also have a go a renaming tar.tar as tar.gz . Your M$ installation might have overridden the file extension to 'help' you.

I hope this helps, and there's plenty to keep you going now.

[Zazicle]

I'm pretty sure that using Python is completely wrong. PyCrypto's implementation of RC5 is only RC5-32/x/x and we need RC5-64 which uses different word (and hence block) sizes.

rccrypt seems to be the only thing we've found that works with RC5-64.

[addybobble]

Looks like you might be right there. AAARGH. I'll have a look at the source and see if I can tweak it. Cygwin should be capable of compiling rccrypt, mind. I'll have a go with that too. It's a shame as I was hoping we could get less techy people on this.

Please bear with me...

[addybobble]

RCCRYPT doesn't build properly in cygwin. I'll need to do some work on it when I get some spare time (I think there is a problem with the handling of 'remainder', plus more fundamental OS specific stuff). This is the version I'm using:

http://www.ricksoft.co.uk/downloads/rccrypt/rccrypt-1.6.tar.gz

[fretty]

Maybe the key is "fivecows" or something along that variation.

[Ashin]

Maybe a little amusement here for anyone that's been chugging at this card...

On the whole five cows theme, I can confirm, in hex, that the password is not:

beefbeefbeefbeef

or

beefbeefbeefbeefbeef (I know it's 10 bytes, was just bored)


As futile as it probably is, I've been letting a computer run for a few weeks now through all the lowercase ascii combinations. Gotten through a and half of b. *sigh*

Also, for anyone who's been at this for a while, after two emails I still haven't heard back yet on whether they used a keyfile, so there's still a chance that ALL of this is in vain. I'll post when I hear back.

[groovygirl20]

had anyone tried cowcowcowcowcow lol

[ajas2]

Hi all,

I'm keen to join the effort, but relcuctant to set my PC off without understanding what it's attempting to do.

I do know what it's *meant* to be doing, but I also want to be able to mess around, try my own theories and combinations, and reassure myself that it's doing what I expect, rather than just running it blind.

I think a lot more people would join in if the attempt was more transparent, and I think the answer is as straightforward as a GUI front-end that people can play with. I've written these things before (albeit not for MS Windows) so I know it's not too tricky. It doesn't need to be pretty either. But I'm a Windows dunce so I can't even try this one.

Hopefully, a really basic front-end shouldn't slow down the processing at all. But if anyone wanted to get really fancy, these are the features I'd love to see:

1) Source code to be read in from a text file. This would allow people to try their own shorter ciphertexts and reassure themselves that things are working as they should. At least we could all see the right ciphertext going in!

2) Option to read trial keys from another text file, to facilitate eg dictionary attacks. (Have we yet tried combinations of words adding up to 8 letters, eg "ThisIsIt")

3) Option to automatically modify the trial keys to attempt uppercase, lowercase sentence case (and others?), as well as what's specificed in the file, all in one go... maybe by a batch process before actually processing the file.

4) Some kind of dynamic readout of results achieved. Writing them all to a CSV or plain text file may be too slow, but how about a series of counters to show the number of ASCII-range characters in the output string, with the highest-scoring ones most likely to contain text. Option to force the program to save all output, eg to use when checking that it's working as expected.

5) Some opportunity to vary the trial parameters, eg the key length is sometimes specified in bits, not bytes. If the key is really only 8 bits then we'd only have 256 to try. I'd try 00001101, being binary 13, or perhaps reverse the bit-order, but then with 256 keys it's trivial to check the lot.

and finally...

6) Not a GUI feature but a request... can someone who knows the parameters we're using at the moment please summarise so that we're all up to speed? As far as I can tell it's: 64-bit blocks (ie 32-bit words), key length 8 bytes, 12 rounds (or is it 6? there's some room for confusion here. The original specs allowed any number, and commented that one RC5 round was equivalent to two DES rounds. Now the specs seem to require even-number-rounds only, so are 12 new rounds equal to 6 old ones?).

If anyone's interested, the original Rivest paper on RC5 is a very easy read in plain English: http://theory.lcs.mit.edu/~rivest/Rivest-rc5rev.pdf

Feel free to ignore my crazy suggestions - I like to wish for the impossible!

Alex.

Webmaster at www.simonsingh.net as it happens, for any of you Code Book fans.

[groovygirl20]

i vote TRASNLTR!

no wait

it exploded

[addybobble]

Hi again.

Thanks for the feedback before about pycrypto: Yes it's the wrong flavour of RC5 (I reviewed the code last night), don't use it. I'm using rccrypt proper on Linux now. People who want to have a go with he windows version should use the ones made available in earlier posts: they look good to me.

Spotted something: After looking though these posts one of the ones from ramsfan mentioned a vertical line going through the card.

I can confirm that on my card it does indeed stroke through the line (from top to bottom) FD 8C A1 39 0B 85 4D BD 88 D7 BA 72 03 A9 11. In fact my line strikes through the right column. This is the line of numbers in line with the leftmost cow.

Could this be an IV (initial vector) we need or a basis for searching for keys? This looks rather unlikely to me as it would be pretty hard to find a grid in that shape (eg search for a grid where, when you run it through rc5 a column matches the initial vector), but it seems to be more than a coincidence to me.

Has anyone else got a line in the same place? These are very accurately printed cards after all...

[groovygirl20]

its a certain batch of silver cards i think : they've got lines down them... mines on riemann and homage in glass but i havent got it on 13th labour .... its nothing to do with the puzzles im afraid

[Hobbess]

no line on mine, just to stop any unnecassary work

[questionboy]

Sorry didn't see this antwhere else

FIVECOWS 8 letters

is it worth a try

[beglee]