Rob Siemborski wrote:
On Tue, 10 Dec 2002, Hack Kampbjorn wrote:


This is on a OpenBSD 3.2, so that would be bmake and flex:
$ flex --version
flex version 2.5.4

Okay, I'm guessing it's more bmake than flex, since the implicit rules
supplied by gmake are creating this file like so:

flex  -t addr-lex.l > addr-lex.c


What should I do instead?

Use gmake? ;)

Seriously though, I highly doubt this is the only place in the code that
there's problems with a non-GNU make.  (Indeed, the same makefile has that
problem with sieve-lex.c/sieve-lex.l).

Or was this really the only problem?
Yes, that's was the only make problem and only when builddir != srcdir.
If you have a complete set of patches, I'm definately willing to look at
them.
Ok, my private OpenBSD port have these patches:
patch-acap_Makefile_in
patch-imap_duplicate_c
patch-imap_sendmail-map_c
patch-imtest_imtest_c
patch-lib_auth_krb_c
patch-lib_auth_krb_pts_c
patch-lib_auth_krb_pts_h
patch-lib_cyrusdb_db3_c
patch-ptclient_ptloader_c
patch-sieve_Makefile_in

The acap and sieve are the ones I already sent.

patch-imtest_imtest_c explicitly includes <openssl/md5.h> as OpenBSD 3.2 comes with OpenSSL 0.9.7-beta3 30 Jul 2002. I suppose this should check if OPENSSL_VERSION_NUMBER >= 0x00907000L
--- imtest/imtest.c.orig Mon Nov 4 17:10:32 2002
+++ imtest/imtest.c Sat Dec 7 15:39:54 2002
@@ -78,6 +78,7 @@

#ifdef HAVE_SSL
#include <openssl/ssl.h>
+#include <openssl/md5.h>

static SSL_CTX *tls_ctx = NULL;
static SSL *tls_conn = NULL;

The rest changes <db.h> to <db3.h> as installed by the db3 port. OpenBSD has the original db1 as <db.h>. The configure script is smart enough to find db3 support with -ldb and <db3.h> but all the cyrus-imapd soruces has hardcoded <db.h>
-Rob

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456
Research Systems Programmer * /usr/contributed Gatekeeper



--
Med venlig hilsen / Kind regards

Hack Kampbjørn

Reply via email to