I have replaced exim-1.90 in the ftp directory with exim-1.91:
ftp://ftp.cus.cam.ac.uk/pub/software/programs/exim/exim-1.91.tar.gz
This is a bug-fix and tidyup release. If you are successfully running
1.90 you may be happy to continue with it. To help you decide, I'll
append the ChangeLog for 1.91 below.
--
Philip Hazel University Computing Service,
ph10@??? New Museums Site, Cambridge CB2 3QG,
P.Hazel@??? England. Phone: +44 1223 334714
1. The extraneous export of EXIMON_EXIM_CONFIG in the eximon script seemed to
cause trouble with some versions of /bin/sh (NetBSD). It seems to have been
there by mistake, so I have removed it, and added a note to the comment about
redefining $config.
2. Recognize os types "aix*" and "AIX*" as AIX.
3. Define T_TXT in dns.c if not defined (SunOS4).
4. Added openbsd*) os=OpenBSD;; to the os-type script.
5. In some filter messages, replaced "log" by "logwrite" (oversight).
6. If an MX pointed to a non-existent host, that host was shown in the -bt
address list with an empty name and no IP address, which was a bit confusing.
Now it shows [unknown] as the address and adds "** unusable **" to the line.
7. If a domainlist router used the bydns option and that produced a list of
hosts, some of which had no IP addresses, those ones were themselves re-
looked up in the DNS as if they were in the router's list; this could cause
looping if any domain names were repeated.
8. Installed PCRE 1.08.
9. scripts/os-type now distinguishes Linux-libc5 as special, retaining the old
configuration files. The main Linux files now assume libc6 and default to using
DB. Also, changed the ps arg for exiwhat to ax instead of -ax for the libc6
(now the default) version.
10. The size of an incoming message, as listed in the S= field in the log line,
was too big - it was including the entire size of the -H file, not just the
contents of the header lines, plus the file name that is put at the start of
the -D file.
11. Count the lines in the body as it is read, and preserve on the spool as
body_linecount; construct a complete message_linecount when reading in the -H
file.
12. The smtp transport now uses message_linecount to give a better estimate of
the size. It adds a fudge factor of 1024, which can be adjusted by a private
option called size_addition.
13. There are occasional esoteric circumstances when Exim tries to write to the
main log when the uid isn't root or exim (e.g. a non-setuid binary with
log_arguments set). Previously it bombed out. Now, if opening the log fails and
the uid is not root or exim, it directs the logging to stderr.
14. Check result of send() when writing SMTP commands, and give correct error
on failure.
15. If a continued delivery down an existing SMTP connection deferred, Exim
went on to try other hosts in an incorrect way, instead of deferring delivery.
This bug was introduced by change 64 of 1.890. Sigh. You mend some, you break
some...
16. The return of fclose() when writing the -H file wasn't being checked -
though it shouldn't fail since prior calls to fflush() and fsync() are checked.
But you never know.
17. Include expansion error text in message generated when added headers fail
to expand.
18. Reply 250 to RSET even when trying to flush a connection after an
unexpected failure.
19. Reply "421 Unexpected failure" to all (non-RSET) commands instead of "503
command out of sequence" when panic-closing with an SMTP channel open.
--
*** Exim information can be found at
http://www.exim.org/ ***
