----- Original Message -----
Sent: Thursday, October 31, 2002 12:45
PM
Subject: Re: Building openssl on
Win2K
I did it, and without any problem worth
mentioning.
Your troubles might be with two things,
though.
One might be the perl configure .. that is needed
to set up the makefile, there is mention of a specific perl distro req'd, I
just ran it with the one I had, and it worked fine (could be the required one,
but I really can't remember which one I installed).
Second is that you might have forgotten to run
vcvars32 before the nmake.
BTW, I built it with VC6 under Win2KPro. There is
also an IDE for VC6, runs just as fine, and as a bonus, compiles all the
openssl tools separately as well.
----- Original Message -----
Sent: Thursday, October 31, 2002 8:09
PM
Subject: Building openssl on
Win2K
Okay, I give up.
I followed
the build instructions in INSTALL.W32 for VC++ only to find an unparseable
makefile (ntdll.mak) with carriage returns embedded in the names of two
macros (e.g. SSL^MOBJ=$(OBJ_D)\ssl.obj ...).
When I fixed that, I
discovered that the makefile was attempting to copy files from the $(SRC_D) (".") directory that actually lived
in its many subdirectories. Rather than perform the major surgery
required to fix that gaff, I decided to fall back, regroup and try plan B,
building under Cygwin.
That got me as far as the first call to
gcc:
gcc -I. -I../include -DTHREADS -DDSO_WIN32 -DTERMIOS
-DL_ENDIAN -fomit-frame-pointer -O2 -m486 -Wall -c -o cryptlib.o
cryptlib.c
cryptlib.c:105: #error "Inconsistency between crypto.h and
cryptlib.c"
cryptlib.c checks for
#if CRYPTO_NUM_LOCKS != 29
#
error "Inconsistency between crypto.h and cryptlib.c"
#endif
Of
course, crypto.h says
#define CRYPTO_NUM_LOCKS 29
but that
doesn't seem to impress cryptlib.c.
At this point I started to get
suspicious...
So my question is - is there anyone who has successfully
built openssl-0.9.6g on any Win32 platform? If so, can I please hear
from you as to how you managed the feat?
Thanks,
-Nick