I'm working on a PowerPC G4 Mac running MacOS X 10.3.9. It has XCode
1.5, including gcc 3.3 with the "November 2004 Update". (I'm not
actually using XCode, but it was the easiest way to get the whole
compiler/linker package installed.) This Mac has to remain at 10.3.9,
so I can't move to a newer XCode or gcc. I will eventually add newer
Macs to the test set, though.
Downloaded openssl-0.9.7m.tar.gz.
Unzipped, untarred with no problems.
"./config threads shared" seemed to work fine. (It guessed
darwin-ppc-cc, which is good: "cc" is equivalent to "gcc" on this
system.)
"make" seemed to work fine. (It failed when I used the "zlib" option,
so I removed that option.)
"make test" spits out SO much data that I can't tell if there's any
errors or not. Some parts of that output present lists of values, but I
have no idea of those values are GOOD or BAD.
Below are places that LOOK like errors to me, but might be
normal/expected for the tests.
This line shows up twice:
16 bytes leaked in 2 chunks
This line shows up twice, near the above line:
error 40 at 0 depth lookup:proxy cerificates not allowed, please
set the appropriate flag
These two lines (with only the first number changing) show up TWENTY-TWO
times:
ERROR in CLIENT
9167:error:1407E086:SSL routines:SSL2_SET_CERTIFICATE:certificate
verify failed:s2_clnt.c:1067:
Here's the last several lines of "make test" output, I think it
describes the environment:
util/shlib_wrap.sh apps/openssl version -a
OpenSSL 0.9.7m 23 Feb 2007
built on: Thu Nov 8 17:23:17 CST 2007
platform: darwin-ppc-cc
options: bn(64,32) md2(int) rc4(ptr,char) des(idx,cisc,16,long)
idea(int) blowfish(ptr)
compiler: cc -DOPENSSL_SYSNAME_MACOSX -fPIC -fno-common -DOPENSSL_PIC
-DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H
-DOPENSSL_NO_KRB5 -O3 -fomit-frame-pointer -DB_ENDIAN
OPENSSLDIR: "/usr/local/ssl"
SO, how can I tell if those apparent error messages in the tests
indicate that my build of OpenSSL is good or bad?? I really, really
have to get OpenSSL setup so I can -use- it on this Mac, but I'm stuck
if the build isn't good.
IN MY OPINION: The "make test" output should be MUCH shorter, ONLY
showing summaries like "test succeeded" or "test failed". A separate
"make testverbose" could give all the extra stuff for isolating and
debugging build problems. There was MUCH less test output when I built
OpenSSL on Win32, and that was very easy to decipher. In fact, I think
it ended with a "all tests good" summary.
Thanks,
Mike
--
Quando Omni Flunkus Moritati
(When all else fails, play dead.)
