I tried building the latest stable snapshot of OpenSSL 0.9.7 on
Tru64 (aka Digital UNIX, aka OSF/1) on a DEC Alpha system.  Output of
"make report" is:

OpenSSL version:  0.9.7e-dev
Last change:      Various fixes to s3_pkt.c so alerts are sent properly....
Options:          --prefix=/usr/users/lcs no-zlib no-threads no-shared
no-krb5
OS (uname):       OSF1 ironwood V5.1 2650 alpha
OS (config):      alpha-dec-tru64
Target (default): alpha-cc
Target:           alpha-cc
Compiler:         Compaq C V6.5-011 on Compaq Tru64 UNIX V5.1B (Rev. 2650)

The attached diffs to the makefile were enough to let it build and
pass tests with "./config no-zlib no-threads no-shared".

I have no problems building 0.9.7 snapshot and can not confirm neither of these:


The specific problems were:
 1. Using the OS's version of libcrypto.a instead of the local one.

[At least in my case] "OS version of libcrypto.a" is 0.9.6. If complier driver were to link with it, I would get get a lot of linking errors. I didn't get any and even double-checking confirmed that it's linked correctly.


 2. Some sketchy shell syntax in makefiles that broke under Tru64's /bin/sh
    easily fixed by adding double-quotes.

What exactly was the problem? Trouble is that proposed change would actually *break* it. With double-quotes it won't advance through "file.1" and "file.2," but will attempt to copy a single file called "file.1 file.2," which doesn't exist.


------------ diffs:
*** ./apps/Makefile.orig Mon Jun 28 18:04:21 2004
--- ./apps/Makefile Thu Jul 1 00:35:14 2004
***************
*** 143,149 ****
$(EXE): progs.h $(E_OBJ) $(PROGRAM).o $(DLIBCRYPTO) $(DLIBSSL)
$(RM) $(EXE)
! if [ "$(SHLIB_TARGET)" = "hpux-shared" -o "$(SHLIB_TARGET)" = "darwin-shared" ] ; then \
$(CC) -o $(EXE) $(CFLAGS) $(PROGRAM).o $(E_OBJ) $(PEX_LIBS) $(DLIBSSL) $(LIBKRB5) $(DLIBCRYPTO) $(EX_LIBS) ; \
else \
LD_LIBRARY_PATH=..:$$LD_LIBRARY_PATH \
--- 143,149 ----
$(EXE): progs.h $(E_OBJ) $(PROGRAM).o $(DLIBCRYPTO) $(DLIBSSL)
$(RM) $(EXE)
! if [ "$(SHLIB_TARGET)" = "hpux-shared" -o "$(SHLIB_TARGET)" = "darwin-shared" -o "$(SHLIB_TARGET)" = "tru64-shared" ] ; then \
$(CC) -o $(EXE) $(CFLAGS) $(PROGRAM).o $(E_OBJ) $(PEX_LIBS) $(DLIBSSL) $(LIBKRB5) $(DLIBCRYPTO) $(EX_LIBS) ; \
else \

If you target no-shared build, this code is never executed... And if so how come you refer to it as explicit problem in no-shared context?


  install:
        @if test -n "$(EXHEADER)"; then \
!         for i in "$(EXHEADER)" ; \

See above.

I'm leaning toward just dismissing this case as so to say groundless... Unless more solid evidence can be presented of something being broken under Tru64 can be presented... A.

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to