procedure to give the correct -I options to the compiler.  This is
*especially* true for test programs that appear in two places, with
different paths to e_os.h depending on where they are built.

doesn't differ between FIPS and non-FIPS modes.

(probably) that FIPS is an entirely internal macro, and is not
accessible by third-party authors.

Update hashes.

str(n)icmp defs).

Disable a few tests if not FIPS.

depending on FIPS mode.

platforms that don't (necessarely) have it.  In the case of VMS, this
means moving a couple of functions from apps/ to crypto/ and make them
general (although only used privately).

0_9_7-stable'.

platforms that don't (necessarely) have it.  In the case of VMS, this
means moving a couple of functions from apps/ to crypto/ and make them
general (although only used privately).

gets confused...  The separate TESTDATA variable was inspired from
crypto/evp/Makefile.ssl.

#if 0 unimplemented ciphers so mkdef.pl doesn't pick
them up.

The current solution is very Unix-bound, and there is probably a better way to do this.

memcmp definition.

update fips_make_sha1 to use fips_err.h

update hashes.

let's change it's suffix from .c to .h.  This also avoids some
otherwise very mysterious (and probably sensible from a historical
point of view :-)) name changes done by mk1mf.pl.

fingerprints.

Otherwise, the test target will fail because libcrypto.sha1 is missing
or not up to date.

that macro.

It's possible that OPENSSL_UNISTD_IO should be used instead of
OPENSSL_UNISTD, for the MSDOS case...