v2.2.28 release candidate released
http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz.sig
Pretty large release. Please test before the final v2.2.28, which should be out in a few days.
BTW. Our plan is to start making new releases approximately every month from now on.
- director: "doveadm director move" to same host now refreshes user's timeout. This allows keeping user constantly in the same backend by just periodically moving the user there.
- When new mailbox is created, use initially INBOX's dovecot.index.cache caching decisions.
- Expunging mails writes GUID to dovecot.index.log now only if the GUID is quickly available from index/cache.
- imap: When BINARY FETCH sees invalid content, return NO [PARSE] reply instead of [UNKNOWNCTE] (which is now used only for actually unknown Content-Transfer-Encoding headers).
- pop3c: Increase timeout for PASS command to 5 minutes.
- Mail access errors are no longer ignored when searching or sorting. With IMAP the untagged SEARCH/SORT reply is still sent the same as before, but NO reply is returned instead of OK.
- Make dovecot.list.index's filename configurable. This is needed when there are multiple namespaces pointing to the same mail root (e.g. lazy_expunge namespace for mdbox).
- Add size.virtual to dovecot.index when folder vsizes are accessed (e.g. quota=count). This is mainly a workaround to avoid slow quota recalculation performance when message sizes get lost from dovecot.index.cache due to corruption or some other reason.
- auth: Support OAUTHBEARER and XOAUTH2 mechanisms. Also support them in lib-dsasl for client side.
- auth: Support filtering by SASL mechanism: passdb { mechanisms }
- Shrink the mail processes' memory usage by not storing settings duplicated unnecessarily many times.
- imap: Add imap_fetch_failure setting to control what happens when FETCH fails for some mails (see example-config).
- imap: Include info about last command in disconnection log line.
- imap: Created new SEARCH=X-MIMEPART extension. It's currently not advertised by default, since it's not fully implemented.
- fts-solr: Add support for basic authentication.
- Cassandra: Support automatically retrying failed queries if execution_retry_interval and execution_retry_times are set.
- doveadm: Added "mailbox path" command.
- mail_log plugin: If plugin { mail_log_cached_only=yes }, log the wanted fields only if it doesn't require opening the email.
- mail_vsize_bg_after_count setting added (see example-config).
- mail_sort_max_read_count setting added (see example-config).
- Index files: day_first_uid wasn't updated correctly since v2.2.26. This caused dovecot.index.cache to be non-optimal.
- imap: SEARCH/SORT may have assert-crashed in client_check_command_hangs
- imap: FETCH X-MAILBOX may have assert-crashed in virtual mailboxes.
- imap: Running time in tagged command reply was often wrongly 0.
- search: Using NOT n:* or NOT UID n:* wasn't handled correctly
- director: doveadm director kick was broken
- director: Fix crash when using director_flush_socket
- director: Fix some bugs when moving users between backends
- imapc: Various error handling fixes and improvements
- master: doveadm process status output had a lot of duplicates.
- autoexpunge: If mailbox's rename timestamp is newer than mail's save-timestamp, use it instead. This is useful when autoexpunging e.g. Trash/* and an entire mailbox is deleted by renaming it under Trash to prevent it from being autoexpunged too early.
- autoexpunge: Multiple processes may have been trying to expunge the same mails simultaneously. This was problematic especially with lazy_expunge plugin.
- auth: %{passdb:*} was empty in auth-worker processes
- auth-policy: hashed_password was always sent empty.
- dict-sql: Merge multiple UPDATEs to a single statement if possible.
- fts-solr: Escape {} chars when sending queries
- fts: fts_autoexpunge_exclude = \Special-use caused crashes
- doveadm-server: Fix leaks and other problems when process is reused for multiple requests (service_count != 1)
- sdbox: Fix assert-crash on mailbox create race
- lda/lmtp: deliver_log_format values weren't entirely correct if Sieve was used. especially %{storage_id} was broken.
On 20 February 2017 at 13:15, Timo Sirainen tss@iki.fi wrote:
http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz.sig
Pretty large release. Please test before the final v2.2.28, which should be out in a few days.
BTW. Our plan is to start making new releases approximately every month from now on.
So by December we'll have had at least 10 new releases? Why do many though? Just curious.
I am running this on two servers already - and monitoring...
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254 7 3200 0004/+254 7 2274 3223 "Oh, the cruft."
On 20 Feb 2017, at 13.43, Odhiambo Washington odhiambo@gmail.com wrote:
BTW. Our plan is to start making new releases approximately every month from now on.
So by December we'll have had at least 10 new releases?
Maybe. Lets see :)
Why do many though? Just curious.
There's a lot of work to do and doesn't make sense to do as huge releases as they now seem to be.
BTW. Related to this: We're now looking for someone to do even more coding work for us, ideally to be located in Finland: https://www.linkedin.com/jobs/view/275257998/ https://www.linkedin.com/jobs/view/275257998/
Hi,
On 20-Feb-17 12:15, Timo Sirainen wrote:
imap: When BINARY FETCH sees invalid content, return NO [PARSE] reply instead of [UNKNOWNCTE] (which is now used only for actually unknown Content-Transfer-Encoding headers).
Has this been tested with Roundcube webmail? I know Roundcube has some workarounds when dovecot now responds with that "[UNKNOWNCTE]".
Best regards, Toni
On 20 Feb 2017, at 13.47, Toni Mattila toni@solu.fi wrote:
Hi,
On 20-Feb-17 12:15, Timo Sirainen wrote:
imap: When BINARY FETCH sees invalid content, return NO [PARSE] reply instead of [UNKNOWNCTE] (which is now used only for actually unknown Content-Transfer-Encoding headers).
Has this been tested with Roundcube webmail? I know Roundcube has some workarounds when dovecot now responds with that "[UNKNOWNCTE]".
No. Hmm. This was one of the changes that I was wondering whether it should be in v2.2 or if it should be left to v2.3. Actually looks like the change was done already last May, so I must have decided earlier not to put it to v2.2, but now semi-accidentally added it.
I'll remove it from the final v2.2.28, but I suppose should tell Roundcube to start preparing for this change in v2.3.
On 20.02.2017 12:47, Toni Mattila wrote:
imap: When BINARY FETCH sees invalid content, return NO [PARSE] reply instead of [UNKNOWNCTE] (which is now used only for actually unknown Content-Transfer-Encoding headers).
Has this been tested with Roundcube webmail? I know Roundcube has some workarounds when dovecot now responds with that "[UNKNOWNCTE]".
I suppose we'll have to fix the Roundcube code where we do:
// handle UNKNOWN-CTE response - RFC 3516, try again with standard BODY request if ($binary && !$found && preg_match('/^' . $key . ' NO \[UNKNOWN-CTE\]/i', $line)) {
Create a ticket for this, please.
In RFC3501 [PARSE] is described as "The human-readable text represents an error in parsing the [RFC-2822] header or [MIME-IMB] headers of a message in the mailbox.". So, is this really an appropriate error code for this case?
-- Aleksander 'A.L.E.C' Machniak Kolab Groupware Developer [http://kolab.org] Roundcube Webmail Developer [http://roundcube.net]
PGP: 19359DC1 # Blog: https://kolabian.wordpress.com
On 20 Feb 2017, at 14.06, A.L.E.C alec@alec.pl wrote:
In RFC3501 [PARSE] is described as "The human-readable text represents an error in parsing the [RFC-2822] header or [MIME-IMB] headers of a message in the mailbox.". So, is this really an appropriate error code for this case?
I think it's the closest to being correct. It's saying that there was some kind of a parsing problem with the email. RFC 3501 only talks about headers, but that's probably because they hadn't really considered parsing errors in email bodies.
The IMAP BINARY RFC only talks about [UNKNOWN-CTE], but it says it should be sent on unknown Content-Transfer-Encoding. So reusing it for other purposes seems more wrong than using [PARSE].
This was discussed some years ago in imap mailing list: http://mailman13.u.washington.edu/pipermail/imap-protocol/2012-December/0019... http://mailman13.u.washington.edu/pipermail/imap-protocol/2012-December/0019... http://mailman13.u.washington.edu/pipermail/imap-protocol/2012-December/0019... http://mailman13.u.washington.edu/pipermail/imap-protocol/2012-December/0019...
On 20/02/2017 10:15, Timo Sirainen wrote:
http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz
I had a minor problem on build. I have configured without compression, [ZFS is providing it]:
./configure
...
--with-bzlib=no
--with-zlib=no
--with-lzma=no
--with-lz4=no \
src/lib-compression/istream-zlib.c uses "#ifdef HAVE_ZLIB" to prevent compiling the compression methods but test-compression.c tries to use these:
Undefined first referenced symbol in file i_stream_create_gz test-compression.o ld: fatal: symbol referencing errors. No output written to test-compression
I avoided the problem by hacking out compression-test from the makefile src/lib-compression/Makefile.in
Otherwise rc1 is working with Solaris 10 / Sun Studio 12.5 - thank you.
Pigeonhole will follow later today.
Op 20-2-2017 om 11:15 schreef Timo Sirainen:
http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz http://dovecot.org/releases/2.2/rc/dovecot-2.2.28.rc1.tar.gz.sig
Pretty large release. Please test before the final v2.2.28, which should be out in a few days.
BTW. Our plan is to start making new releases approximately every month from now on.
- director: "doveadm director move" to same host now refreshes user's timeout. This allows keeping user constantly in the same backend by just periodically moving the user there.
- When new mailbox is created, use initially INBOX's dovecot.index.cache caching decisions.
- Expunging mails writes GUID to dovecot.index.log now only if the GUID is quickly available from index/cache.
- imap: When BINARY FETCH sees invalid content, return NO [PARSE] reply instead of [UNKNOWNCTE] (which is now used only for actually unknown Content-Transfer-Encoding headers).
- pop3c: Increase timeout for PASS command to 5 minutes.
- Mail access errors are no longer ignored when searching or sorting. With IMAP the untagged SEARCH/SORT reply is still sent the same as before, but NO reply is returned instead of OK.
- Make dovecot.list.index's filename configurable. This is needed when there are multiple namespaces pointing to the same mail root (e.g. lazy_expunge namespace for mdbox).
- Add size.virtual to dovecot.index when folder vsizes are accessed (e.g. quota=count). This is mainly a workaround to avoid slow quota recalculation performance when message sizes get lost from dovecot.index.cache due to corruption or some other reason.
- auth: Support OAUTHBEARER and XOAUTH2 mechanisms. Also support them in lib-dsasl for client side.
- auth: Support filtering by SASL mechanism: passdb { mechanisms }
- Shrink the mail processes' memory usage by not storing settings duplicated unnecessarily many times.
- imap: Add imap_fetch_failure setting to control what happens when FETCH fails for some mails (see example-config).
- imap: Include info about last command in disconnection log line.
- imap: Created new SEARCH=X-MIMEPART extension. It's currently not advertised by default, since it's not fully implemented.
- fts-solr: Add support for basic authentication.
- Cassandra: Support automatically retrying failed queries if execution_retry_interval and execution_retry_times are set.
- doveadm: Added "mailbox path" command.
- mail_log plugin: If plugin { mail_log_cached_only=yes }, log the wanted fields only if it doesn't require opening the email.
- mail_vsize_bg_after_count setting added (see example-config).
- mail_sort_max_read_count setting added (see example-config).
- Index files: day_first_uid wasn't updated correctly since v2.2.26. This caused dovecot.index.cache to be non-optimal.
- imap: SEARCH/SORT may have assert-crashed in client_check_command_hangs
- imap: FETCH X-MAILBOX may have assert-crashed in virtual mailboxes.
- imap: Running time in tagged command reply was often wrongly 0.
- search: Using NOT n:* or NOT UID n:* wasn't handled correctly
- director: doveadm director kick was broken
- director: Fix crash when using director_flush_socket
- director: Fix some bugs when moving users between backends
- imapc: Various error handling fixes and improvements
- master: doveadm process status output had a lot of duplicates.
- autoexpunge: If mailbox's rename timestamp is newer than mail's save-timestamp, use it instead. This is useful when autoexpunging e.g. Trash/* and an entire mailbox is deleted by renaming it under Trash to prevent it from being autoexpunged too early.
- autoexpunge: Multiple processes may have been trying to expunge the same mails simultaneously. This was problematic especially with lazy_expunge plugin.
- auth: %{passdb:*} was empty in auth-worker processes
- auth-policy: hashed_password was always sent empty.
- dict-sql: Merge multiple UPDATEs to a single statement if possible.
- fts-solr: Escape {} chars when sending queries
- fts: fts_autoexpunge_exclude = \Special-use caused crashes
- doveadm-server: Fix leaks and other problems when process is reused for multiple requests (service_count != 1)
- sdbox: Fix assert-crash on mailbox create race
- lda/lmtp: deliver_log_format values weren't entirely correct if Sieve was used. especially %{storage_id} was broken.
On Mon Feb 20 2017 05:15:34 GMT-0500 (Eastern Standard Time), Timo Sirainen tss@iki.fi wrote:
- auth: Support OAUTHBEARER and XOAUTH2 mechanisms. Also support them in lib-dsasl for client side.
So... does this mean dovecot now has OAUTH2 support? If so... yay! and I'll go open a Thunderbird bug to add support for dovecot's OAUTH2...
On 20.02.2017 23:56, Tanstaafl wrote:
On Mon Feb 20 2017 05:15:34 GMT-0500 (Eastern Standard Time), Timo Sirainen tss@iki.fi wrote:
- auth: Support OAUTHBEARER and XOAUTH2 mechanisms. Also support them in lib-dsasl for client side. So... does this mean dovecot now has OAUTH2 support? If so... yay! and I'll go open a Thunderbird bug to add support for dovecot's OAUTH2...
Yes. It support OAUTHBEARER and XOAUTH2 bearer authentication. See https://wiki.dovecot.org/PasswordDatabase/oauth2
Aki
participants (8)
-
A.L.E.C
-
Aki Tuomi
-
James
-
Odhiambo Washington
-
Stephan Bosch
-
Tanstaafl
-
Timo Sirainen
-
Toni Mattila