On 2013-04-10 18:28:04, Bart Martens wrote: > block 704464 by 704476 > block 704464 by 704474 > stop > > On Wed, Apr 10, 2013 at 04:46:28PM +0200, Ansgar Burchardt wrote: > > Hi, > > > > is there any reason to use a pure Python implementation for AES instead > > of one written with a C backend? > > > > The idea of slowaes seems similar to Digest::SHA::PurePerl where several > > people mentioned it's not too useful on Debian[1]. > > [1] <http://bugs.debian.org/556386> > > It's a dependency of electrum. > http://bitcoin.org/en/choose-your-wallet > http://electrum.org/ > http://bugs.debian.org/704464
I'd rather see electrum ported to one of the many providers of AES for Python that we already have in Debian. python-keyczar, python-m2crypto, and python-crypto come to my mind and I'm probably missing others. Especially for an implementation of an crypto algorithm I'd expect it to ship a better test suite than what slowaes provides. Just running random data through encrypt and decrypt and checking if it matches with the input is just not enough to convince me that the AES implementation is actually correct. [1] Regards [1] From what I can see in the test suite, the FIPS test vectors are only run for the JS implementation, but not for the Python one. Please correct me if I'm just missing something. -- Sebastian Ramacher
signature.asc
Description: Digital signature
