Fix or replace Crypto++

[GoogleCodeExporter]

There was always something wrong with our install of Crypto++; however, I 
hacked around it with a global pointer. I'm not in the least surprised that 
MinGW didn't like that very much.

We need to either fix Crypto++ or replace it with, say, OpenSSL. All we need is 
a digest algorithm (MD5 is what I'd prefer, if possible). Might take some 
doing, but shouldn't be too taxing overall. 

Original issue reported on code.google.com by seth.h...@gmail.com on 13 Jan 2011 at 6:11

• Blocking: #137

[GoogleCodeExporter]

Note that, although OpenSSL is huge, its MD5 includes consist only of the 
following 3 files:

e_os2.h
opensslconf.h
md5.h

So if the *.lib's small, I'm leaning this way.

By the way, getting a digest with OpenSSL seems simple enough:
http://stackoverflow.com/questions/1220046/in-c-how-to-get-md5-hash-of-a-file/12
20177#1220177

Original comment by seth.h...@gmail.com on 13 Jan 2011 at 8:10

[GoogleCodeExporter]

Side-note: OpenSSL has a somewhat weird license. (It's Apache 1 + BSD). Apache 
1 required a citation --which I don't mind at all, but at least led me to look 
briefly for more options. L Peter Deutsch's MD5 library stood out:
http://sourceforge.net/project/showfiles.php?group_id=42360

It's based on the original RFC and is only 3 files in size. The license is 
friendly. 
In addition, I trust Deutsch's work (if it is, indeed, the same guy from 
Aladdin software who wrote Ghostscript). 

Worth trying out, checking performance, etc.

Original comment by seth.h...@gmail.com on 13 Jan 2011 at 8:10

[GoogleCodeExporter]

Fixed; used the tiny MD5 library. 

Original comment by seth.h...@gmail.com on 13 Jan 2011 at 10:31

• Changed state: Fixed