Also, we're concerned about the volume of preprocessor directives this patch inserts into the middle of functions. Would it be possible to define the CMS symbols in terms of the PKCS7 symbols for OpenSSL pre-1.0?