Re: [Zsync-users] Considering to generate .zsync files in MirrorBrain

[Colin P. <cp...@mo...>]

Peter Pöml wrote:
> Hi Colin,
> 
> Am 22.03.2010 um 20:47 schrieb Colin Phipps:
>> I just wrote a unit test for librcksum/md4.c, using one of the test
>> vectors from RFC1320 , and it seems fine - it agrees with the RFC and
>> openssl(1).
> 
> Thank you for the test program. I checked it and the result I got was indeed fine.
> 
> Now, the next thing I suspected was that Python was handing over the data to OpenSSL wrongly. However, when searching the Internet, I found a usage example on http://barryp.org/software/py-md4/ that immediately raised my suspicion that it I used the Python hashlib module in the wrong way. In fact, I incorrectly passed a string to the MD4 context upon its creation that changed the hash in effect. 
> 
> I used hashlib.new('md4', string='md4') where I should have used hashlib.new('md4'). I stupidly assumed that the second argument was required and effectively chose the algorithm.
> 
> I'm really sorry for having wasted your time! I'm grateful for your help.

No problem :-). Since I had managed to get the MD4 wrong once myself, it
was worth adding a unit test for.

-- 
Colin Phipps <cp...@mo...>