On Jan 24, 2014, at 5:58, Tim Burgess <[email protected]> wrote: >> OK, >> I've disabled all of the export statements in my .bash_profile and performed >> the following tests using the default gnumakefile and code extracted form >> cryptopp562 instead of SVN: >> 1 - Using the default STD library: >> make static test VERBOSE=1 >> This builds successfully (with a few warnings) and passes all tests. The log >> is at https://dl.dropboxusercontent.com/u/29578720/Crypto-no_export.txt
From what I see, this produced a working library. You should be able to use it. I would stop here. You may add "-maes" and other flags if you wish. What more do you need (and why? :-)? >> 2 - Using the libc++ STD library: >> make static test VERBOSE=1 -stdlib=libc >> Again, this claims to build all targets successfully, but the .o and .a >> files are all 0-length. The log is at >> https://dl.dropboxusercontent.com/u/29578720/Crypto-no_export_libc%2B%2B.txt How much more proof do you need that this is not a good idea? >> Unfortunately, I need to use libc++, as other components in my application >> project require it. Any ideas for my next step, assuming that I don't have a >> problem with applying patches, per se. See if you can do the same as your first test with my makefile. Or just leave it alone: you got a working library, now install it to where it (and the include files) should live in your system, compile your code with the same compiler, and link to it. You may (or may not) lose on performance, but otherwise it should work.
smime.p7s
Description: S/MIME cryptographic signature
