On 1/2/07, Ruediger Pluem <[EMAIL PROTECTED]> wrote:
Can we be sure that M5_LONG is always 32bit on all platforms that are supported by apr-util?
Well, probably. Ideally, the 'right' solution would be to use their structure rather than shoe-horning it into our slightly different structure. For a proof of concept, yah, it works for me. =)
But wouldn't this also solve our possible license problems with RSA regarding MD5, which is currently investigated?
Yes, or rather it would make the problem OpenSSL's and not ours.
Plus it would relief us from the ugly hack above. But regardles of this, would we be able to duplicate their optimized implementation without getting any copyright trouble with the openssl guys?
The OpenSSL optimized MD5 implementation is based on code written here: http://etud.epita.fr/~bevand_m/papers/md5-amd64.html So, the AMD64 assembly is under public domain (and a valid disclaim of copyright, so actually is public domain). The OpenSSL folks have done some tweaking to the assembly to make it more portable. Plus, OpenSSL is under a modified BSD license, so we can incorporate their code as long as we attribute properly: http://www.openssl.org/source/license.html But, again, really, the 'right' thing is to figure out to marshal into their API rather than duplicate their implementation. I'm fine with an unoptimized fallback, but if OpenSSL is available and they've done the hard work to optimize it, I'd like APR-using programs to benefit. -- justin
