diff options
Diffstat (limited to 'usr.sbin/sendmail/RELEASE_NOTES')
| -rw-r--r-- | usr.sbin/sendmail/RELEASE_NOTES | 141 |
1 files changed, 140 insertions, 1 deletions
diff --git a/usr.sbin/sendmail/RELEASE_NOTES b/usr.sbin/sendmail/RELEASE_NOTES index 9f4f3abd54f..b3e1a8b33d3 100644 --- a/usr.sbin/sendmail/RELEASE_NOTES +++ b/usr.sbin/sendmail/RELEASE_NOTES @@ -1,11 +1,150 @@ SENDMAIL RELEASE NOTES - @(#)RELEASE_NOTES 8.8.7.7 (Berkeley) 8/3/97 + @(#)RELEASE_NOTES 8.8.8.4 (Berkeley) 10/24/97 This listing shows the version of the sendmail binary, the version of the sendmail configuration files, the date of release, and a summary of the changes in that release. +8.8.8/8.8.8 97/10/24 + If the check_relay ruleset failed, the relay= field was logged + incorrectly. Problem noted by Kari Hurtta of the Finnish + Meteorological Institute. + If /usr/tmp/dead.letter already existed, sendmail could not + add additional bounces to it. Problem noted by Thomas J. + Arseneault of SRI International. + If an SMTP mailer used a non-standard port number for the outgoing + connection, it would be displayed incorrectly in verbose mode. + Problem noted by John Kennedy of Cal State University, Chico. + Log the ETRN parameter specified by the client before altering them + to internal form. Suggested by Bob Kupiec of GES-Verio. + EXPN and VRFY SMTP commands on malformed addresses were logging as + User unknown with bogus delay= values. Change them to log + the same as compliant addresses. Problem noted by Kari E. + Hurtta of the Finnish Meteorological Institute. + Ignore the debug resolver option unless using sendmail debug trace + option for resolver. Problem noted by Greg Nichols of Wind + River Systems. + If SingleThreadDelivery was enabled and the remote server returned a + protocol error on the DATA command, the connection would be + closed but the persistent host status file would not be + unlocked so other sendmail processes could not deliver to + that host. Problem noted by Peter Wemm of DIALix. + If queueing up a message due to an expensive mailer, don't increment + the number of delivery attempts or set the last delivery + attempt time so the message will be delivered on the next + queue run regardless of MinQueueAge. Problem noted by + Brian J. Coan of the Institute for Global Communications. + Authentication warnings of "Processed from queue _directory_" and + "Processed by _username_ with -C _filename_" would be logged + with the incorrect timestamp. Problem noted by Kari E. Hurtta + of the Finnish Meteorological Institute. + Use a better heuristic for detecting GDBM. + Log null connections on dropped connections. Problem noted by + Jon Lewis of Florida Digital Turnpike. + If class dbm maps are rebuilt, sendmail will now detect this and + reopen the map. Previously, they could give stale + results during a single message processing (but would + recover when the next message was received). Fix from + Joe Pruett of Q7 Enterprises. + Do not log failures such as "User unknown" on -bv or SMTP VRFY + requests. Problem noted by Kari E. Hurtta of the + Finnish Meteorological Institute. + Do not send a bounce message back to the sender regarding bad + recipients if the SMTP connection is dropped before the + message is accepted. Problem noted by Kari E. Hurtta of the + Finnish Meteorological Institute. + Use "localhost" instead of "[UNIX: localhost]" when connecting to + sendmail via a UNIX pipe. This will allow rulesets using + $&{client_name} to process without sending the string through + dequote. Problem noted by Alan Barrett of Internet Africa. + A combination of deferred delivery mode, a double bounce situation, + and the inability to save a bounce message to + /var/tmp/dead.letter would cause sendmail to send a bounce + to postmaster but not remove the offending envelope from the + queue causing it to create a new bounce message each time the + queue was run. Problem noted by Brad Doctor of Net Daemons + Associates. + Remove newlines from hostname information returned via DNS. There are + no known security implications of newlines in hostnames as + sendmail filters newlines in all vital areas; however, this + could cause confusing error messages. + Starting with sendmail 8.8.6, mail sent with the '-t' option would be + rejected if any of the specified addresses were bad. This + behavior was modified to only reject the bad addresses and not + the entire message. Problem noted by Jozsef Hollosi of + SuperNet, Inc. + Use Timeout.fileopen when delivering mail to a file. Suggested by + Bryan Costales of InfoBeat, Inc. + Display the proper Final-Recipient on DSN messages for non-SMTP + mailers. Problem noted by Kari E. Hurtta of the + Finnish Meteorological Institute. + An error in calculating the available space in the list of addresses + for logging deliveries could cause an address to be silently + dropped. + Include the initial user environment if sendmail is restarted via + a HUP signal. This will give room for the process title. + Problem noted by Jon Lewis of Florida Digital Turnpike. + Mail could be delivered without a body if the machine does not + support flock locking and runs out of processes during + delivery. Fix from Chuck Lever of the University of Michigan. + Drop recipient address from 251 and 551 SMTP responses per RFC 821. + Problem noted by Kari E. Hurtta of the Finnish Meteorological + Institute. + Make sure non-rebuildable database maps are opened before the + rebuildable maps (i.e. alias files) in case the database maps + are needed for verifying the left hand side of the aliases. + Problem noted by Lloyd Parkes of Victoria University. + Make sure sender RFC822 source route addresses are alias expanded for + bounce messages. Problem noted by Juergen Georgi of + RUS University of Stuttgart. + Minor lint fixes. + Return a temporary error instead of a permanent error if an LDAP map + search returns an error. This will allow sequenced maps which + use other LDAP servers to be checked. Fix from Booker Bense + of Stanford University. + When automatically converting from quoted printable to 8bit text do + not pad bare linefeeds with a space. Problem noted by Theo + Nolte of the University of Technology Aachen, Germany. + Portability: + Non-standard C compilers may have had a problem compiling + conf.c due to a standard C external declaration of + setproctitle(). Problem noted by Ted Roberts of + Electronic Data Systems. + AUX: has a broken O_EXCL implementation. Reported by Jim + Jagielski of jaguNET Access Services. + BSD/OS: didn't compile if HASSETUSERCONTEXT was defined. + Digital UNIX: Digital UNIX (and possibly others) moves + loader environment variables into the loader memory + area. If one of these environment variables (such as + LD_LIBRARY_PATH) was the last environment variable, + an invalid memory address would be used by the process + title routine causing memory corruption. Problem + noted by Sam Hartman of Mesa Internet Systems. + GNU libc: uses an enum for _PC_CHOWN_RESTRICTED which caused + chownsafe() to always return 0 even if the OS does + not permit file giveaways. Problem noted by + Yasutaka Sumi of The University of Tokyo. + IRIX6: Syslog buffer size set to 512 bytes. Reported by + Gerald Rinske of Siemens Business Services VAS. + Linux: Pad process title with NULLs. Problem noted by + Jon Lewis of Florida Digital Turnpike. + SCO OpenServer 5.0: SIOCGIFCONF ioctl call returns an + incorrect value for the number of interfaces. + Problem noted by Chris Loelke of JetStream Internet + Services. + SINIX: Update for Makefile and syslog buffer size from Gerald + Rinske of Siemens Business Services VAS. + Solaris: Make sure HASGETUSERSHELL setting for SunOS is not + used on a Solaris machine. Problem noted by + Stephen Ma of Jtec Pty Limited. + CONFIG: SINIX: Update from Gerald Rinske of Siemens Business + Services VAS. + MAKEMAP: Use a better heuristic for detecting GDBM. + CONTRIB: expn.pl: Updated version from the author, David Muir Sharnoff. + OP.ME: Document the F=i mailer flag. Problem noted by Per Hedeland of + Ericsson. + 8.8.7/8.8.7 97/08/03 If using Berkeley DB on systems without O_EXLOCK (open a file with an exclusive lock already set -- i.e., almost all systems |