path: root/CHANGES.auth
diff options
authorJohn Denker <>2012-06-03 03:47:50 (GMT)
committerJohn Denker <>2012-06-03 03:47:50 (GMT)
commitb8caf78e0ae86118cd0659e28da919721106eadb (patch)
tree62e6afbb7b848a5ac67326ac61d04c6b59ab0251 /CHANGES.auth
parent6116a98a0ec20c22d8e18fcaf384bad18fc39a00 (diff)
bring a few odds and ends into the repo
Diffstat (limited to 'CHANGES.auth')
1 files changed, 62 insertions, 0 deletions
diff --git a/CHANGES.auth b/CHANGES.auth
new file mode 100644
index 0000000..80c8935
--- /dev/null
+++ b/CHANGES.auth
@@ -0,0 +1,62 @@
+- initial release
+- wrong condition in analysis of response. fixed
+- unnecesary debug output freezed eudora. now it should work.
+- added auth state support as described in rfc 2554. It prevents you
+from authenticated again, when you are already authenticated.
+- fixed few error codes to conform with rfc
+- errors during authentication does not end the smtp connection anymore as
+rfc requires.
+- second cram-md5 authentication after failure of first resulted in wrong
+challenge generation (in all subsequent authentications too)
+- major bug in login implementation fixed (bug introduced in 0.21). it
+should work again
+- added a more clean netscape compatibility fix (documented in readme)
+- added the new style (rfc recommended) greeting message. Yoo can select
+beetwen them by #defining or #undefining USE_NEW_GREETING and
+USE_OLD_GREETING on the begining of qmail-smtpd.c. You can even enable both
+to maintain highest degree of compatibility with various clients. This fix
+was suggested by David Harris <>, the developer of
+Pegasus Mail.
+- cram-md5 fix. I wonder, how it ever worked for me. 1 week of my beta
+tests. 3 months of tests on our mail server and the fundamental bug in
+cram-md5 has been found so long after 0.23 release? Anyway it should work
+- changed the error message when problems with starting checkpassword
+occour. This was a big source of question ("out of memory") that i receive
+in my e-mail.
+- Joerg Strohmayer <> showed me light in the dark. I've
+read the SMTP AUTH RFC and assumend that if it uses SASL'a PLAIN, it has to
+be implemented exactly like SASL RFC says. Poor me... SMTP AUTH says that it
+also has to be base64 encoded. This is the relief for the infamous
+NETSCAPE_WORKAROUND. Again big thanks for Joerg for not only telling me
+where the problem is but also for providing the fully working patch.
+- security fix - if your checkpassword crashed, attacker would get the relay
+permission. Now it should work. Reported by Javier Kohen <>,
+thank you!
+- codebase switch. From now on have merged efforts with Eric M. Johnston
+<> so now there are no more 2 separate auth patches but
+just one.
+- arguments to qmail-smtpd have changed! See README.auth for further info.
+- the code is djbized right now
+- TCPREMOTEINFO is set if sender is properly authenticated.
+- bug: AUTH PLAIN 334 response not RFC compliant. Reported by Mark Crispin
+ <mrc@CAC.Washington.EDU>.
+- change: Set TCPREMOTEINFO environment variable to authenticated username.
+ (Previously only set locally to qmail-smtpd.)