Philip Hazel [Thu, 27 Jan 2005 15:15:30 +0000 (15:15 +0000)]
Forgot to mention in the last commit message: Ensure that stdin, stdout,
and stderr are connected to /dev/null when the daemon is running in the
background.
Philip Hazel [Fri, 14 Jan 2005 11:06:58 +0000 (11:06 +0000)]
Incorporated the 4.44 ChangeLog into the main ChangeLog so that a
historical record is preserved. It is separate from the 4.50 entries,
many of which are duplicates, and there's an explanation about why.
Philip Hazel [Thu, 13 Jan 2005 11:12:12 +0000 (11:12 +0000)]
The patch to support LDAP_SEARCH_RES_REFERENCE was unconditional; it
seems that some versions of LDAP don't have this (e.g. OpenLDAP v1, which is
of course pretty old, but some people still use it). I've modified the
patch to exclude the functionality when the macro isn't defined.
Philip Hazel [Thu, 13 Jan 2005 10:09:36 +0000 (10:09 +0000)]
(1) Get rid of compiler warnings (a) in expand.c on 64-bit systems (b) in
malware.c where there was a %hu and an int variable in sscanf(). (2)
Remove an email address from David Saez's email address in malware.c
(best not to have emails in public code these days). (3) Added a credit
to David S to ACKNOWLEDGMENTS.
Philip Hazel [Wed, 12 Jan 2005 12:24:13 +0000 (12:24 +0000)]
(1) The flags in acl.c for fakereject allowed it in too many places. It
is now allowed only when receiving a message. (2) There was a redundant
setting of fake_reject=FALSE in spool_in.c.
Philip Hazel [Tue, 11 Jan 2005 15:17:51 +0000 (15:17 +0000)]
Updated filter source; preliminary update of ACL chapter and addition of
new content scanning chapter in the main doc source; updated
ACKNOWLEDGMENTS from Tom's list for the virus scanners.
Philip Hazel [Wed, 29 Dec 2004 16:34:31 +0000 (16:34 +0000)]
The various Exiscan tests were locked out from the non-SMTP ACL, but
there is no reason why they should not be available (and they apparently
were available in the original patch).
Philip Hazel [Wed, 29 Dec 2004 16:24:03 +0000 (16:24 +0000)]
(1) Setting WITH_OLD_DEMIME without WITH_CONTENT_SCAN caused eximon not
to compile; WITH_CONTENT_SCAN was forced in exim.h in this case, but was
not forced for eximon. I have updated buildconfig so that it gets set in
config.h and thus applies everywhere. (2) Fixed some compiler warnings
in the demime.c module. (3) Moved the inclusion of config.h above the
inclusion of mytypes.h and macros.h for eximon, to be the same as for
exim itself.
Philip Hazel [Wed, 29 Dec 2004 10:16:52 +0000 (10:16 +0000)]
The host_aton() buffer overflow: (1) Put a check in host_aton() itself;
(2) noted that the exploit via dnsdb/ptr lookup was already fortuitously
fixed by a previous change.
Philip Hazel [Tue, 21 Dec 2004 14:38:02 +0000 (14:38 +0000)]
(1) Doubled size of debug_buffer. (2) Output from SpamAssassin test
changed because 96 hours have passed. Shows up the difficulty of testing
such things...
Philip Hazel [Thu, 25 Nov 2004 15:29:36 +0000 (15:29 +0000)]
OpenSSL: make same change as recently for GnuTLS: show file name if
certificate or key file setup fails, because the message from the
library doesn't. Also, update all the OpenSSL TLS tests, which hadn't
been tried for a while, for various recent changes such as protocol
names.
Philip Hazel [Wed, 17 Nov 2004 16:12:26 +0000 (16:12 +0000)]
Allow both strings to be omitted in "${if" expansions: the true value
defaults to "true", which works nicely for "condition" conditions (the
false value has always defaulted to "").
Philip Hazel [Fri, 12 Nov 2004 16:54:55 +0000 (16:54 +0000)]
Exim went into a mad DNS lookup loop when doing a callout where the
host was specified on the transport, if the DNS lookup yielded more than
one IP address.
Philip Hazel [Fri, 12 Nov 2004 15:03:40 +0000 (15:03 +0000)]
(a) Allow an empty sender to be matched against a lookup in an address list.
Previously the only cases considered were a regular expression, or an
empty pattern.
(b) Fixed typo in the FAQ ("prefix" should have been "local_part_prefix").
Philip Hazel [Fri, 12 Nov 2004 12:01:52 +0000 (12:01 +0000)]
Installed the latest exipick (20041110) from John Jetmore, with changes
as follows:
* 20041102 added optimization when searching for specific message_id
* 20041103 changed '=' op to be caseless by dafault, added --caseful
option to change this back to old behavior.
* 20041110 fixed recipients_del POD to note that it includes generated
addrs
* 20041110 added variables recipients_{del,undel}_count