[Dovecot] 1.0 RC1 status #2
I've done several fixes since my original RC1 status mail.
Most importantly I noticed that SEARCH command has been somewhat broken for a long time if the search condition contained OR or NOT conditions. Then again fixing that caused Dovecot to assert-crash for some searches. That should be fixed now. Hopefully I didn't break the SEARCH in other ways while doing this..
So, again, please test the latest CVS or the latest nightly snapshot from http://dovecot.org/nightly/
Especially I'm interested in:
- Does SEARCH work correctly?
- Does ioloop checks now work properly? There were mails about it being broken with kqueue and epoll.
- Does mboxes with CR+LF linefeeds work correctly?
I'm hoping to release the 1.0 RC1 late in sunday or in monday morning (EEST).
Then a status update of my life in case you're interested:
I got in pretty much everywhere I applied for: CS and biotech in Helsinki university and biotech in Tampere university. For Helsinki's biochem I'm the first in the waiting list, which means I almost certainly would get in there too.
So, unless something suddenly changes my mind, I'm moving to Tampere in two months. I'm selling my current apartment, so now's your chance to live in the same place where most of Dovecot was designed and written ;) Or I think it could be a good investment also if you have the money. http://www.igglo.fi/ad.php?card_id=82322 has some information (in Finnish only), I'll get pictures from inside tomorrow. Would be nice if I didn't have to pay real estate agents (?) to sell it for me, so please buy it :)
I'll keep my current job so this shouldn't affect the commercial side of Dovecot.
On Sun, 2006-06-25 at 01:15 +0300, Timo Sirainen wrote: <snip>
Then a status update of my life in case you're interested:
I got in pretty much everywhere I applied for: CS and biotech in Helsinki university and biotech in Tampere university. For Helsinki's biochem I'm the first in the waiting list, which means I almost certainly would get in there too.
Hey Timo, congratulations! Good luck with your bichem/biotech studies. IMHO, molecular biology is a really interesting field but it'll make your head spin if you don't keep up with the pace of the technology. Do yourself a favor and volunteer in a suitable professor's lab so you get a good idea of life in the fast lane ;-)
Stephen
Quoting Timo Sirainen tss@iki.fi:
I've done several fixes since my original RC1 status mail.
Most importantly I noticed that SEARCH command has been somewhat broken for a long time if the search condition contained OR or NOT conditions. Then again fixing that caused Dovecot to assert-crash for some searches. That should be fixed now. Hopefully I didn't break the SEARCH in other ways while doing this..
So, again, please test the latest CVS or the latest nightly snapshot from http://dovecot.org/nightly/
Especially I'm interested in:
- Does SEARCH work correctly?
- Does ioloop checks now work properly? There were mails about it being broken with kqueue and epoll.
- Does mboxes with CR+LF linefeeds work correctly?
I'm hoping to release the 1.0 RC1 late in sunday or in monday morning (EEST).
Then a status update of my life in case you're interested:
I got in pretty much everywhere I applied for: CS and biotech in Helsinki university and biotech in Tampere university. For Helsinki's biochem I'm the first in the waiting list, which means I almost certainly would get in there too.
Congratulations Timo!
So, unless something suddenly changes my mind, I'm moving to Tampere in two months. I'm selling my current apartment, so now's your chance to live in the same place where most of Dovecot was designed and written ;) Or I think it could be a good investment also if you have the money. http://www.igglo.fi/ad.php?card_id=82322 has some information (in Finnish only), I'll get pictures from inside tomorrow. Would be nice if I didn't have to pay real estate agents (?) to sell it for me, so please buy it :)
I'll keep my current job so this shouldn't affect the commercial side of Dovecot.
-- panic: kernel trap (ignored)
FreeBSD 5.4-RELEASE-p12 (SMP - 900x2) Tue Mar 7 19:37:23 PST 2006 /////////////////////////////////////////////////////////////////
Hi Timo,
On 25/06/2006 10:15 a.m., Timo Sirainen wrote:
I've done several fixes since my original RC1 status mail.
Most importantly I noticed that SEARCH command has been somewhat broken for a long time if the search condition contained OR or NOT conditions. Then again fixing that caused Dovecot to assert-crash for some searches. That should be fixed now. Hopefully I didn't break the SEARCH in other ways while doing this..
So, again, please test the latest CVS or the latest nightly snapshot from http://dovecot.org/nightly/
Especially I'm interested in:
- Does SEARCH work correctly?
- Does ioloop checks now work properly? There were mails about it being broken with kqueue and epoll.
Using --with-ioloop=epoll used to work but now I'm seeing this in config.log:
configure:21880: gcc -o conftest -std=gnu99 -g -Wall -W -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar-subscripts -Wformat=2 -Wbad-function-cast conftest.c >&5 conftest.c:72: error: expected declaration specifiers or '...' before 'epoll_create' conftest.c:72: error: expected declaration specifiers or '...' before 'size' conftest.c:74: warning: return type defaults to 'int' conftest.c:74: warning: no previous prototype for '_syscall1' conftest.c: In function '_syscall1': conftest.c:75: error: expected '=', ',', ';', 'asm' or '__attribute__' before '{' token conftest.c:72: error: parameter name omitted conftest.c:72: error: parameter name omitted conftest.c:77: error: expected '{' at end of input configure:21883: $? = 1 configure: program exited with status 1 configure: failed program was: | /* confdefs.h. */ | | #define PACKAGE_NAME "dovecot" | #define PACKAGE_TARNAME "dovecot" | #define PACKAGE_VERSION "1.0.beta9"
<snip>
| int main() | { | return epoll_create(5) < 1; | } | configure:21907: WARNING: epoll ioloop requested but epoll_create() is not available configure:22048: checking if we can use inotify
The system is a Fedora devel (FC6 test1) release on x86_64 with dovecot-cvs/branch. Not sure when it started happening but I can try some other older versions if that's any use in narrowing it all down.
Congratulations on your progress with getting into University.
reuben
Timo Sirainen tss@iki.fi writes:
I'll keep my current job so this shouldn't affect the commercial side of Dovecot.
OK, I'll bite--what is the "commercial side of Dovecot"?
You're such a mystery! :)
On Mon, 2006-06-26 at 09:57 -0400, Mark Plaksin wrote:
Timo Sirainen tss@iki.fi writes:
I'll keep my current job so this shouldn't affect the commercial side of Dovecot.
OK, I'll bite--what is the "commercial side of Dovecot"?
It just means that it's still possible to pay me to develop new features for Dovecot, or to pay for support. There are a couple of companies who have been doing that for a while.
That reminds me that I was going to add "Credits" list to dovecot.org's main page a while ago. Added now. The doc/AUTHORS file is now also linked in there. If you think you belong in there, send me a patch :)
Timo Sirainen wrote:
- Does ioloop checks now work properly? There were mails about it being broken with kqueue and epoll.
About epoll, I only read this once and the poster didn't answer when I asked what is wrong with it. I use epoll and had no single problem with it since the last fix.
but something else:
rawlog in:
3 namespace 4 lsub "" "*" 5 list "" "INBOX" 6 select "ml.denic" 7 UID fetch 1:* (FLAGS) 8 UID fetch 1:1110,1113:1155 (UID RFC822.SIZE FLAGS BODY.PEEK[HEADER.FIELDS (From To Cc Subject Date Message-ID Priority X-Priority References Newsgroups In-Reply-To Content-Type)])
gives normally:
- OK [RAWLOG TIMESTAMP] 2006-06-27 12:39:46
- NAMESPACE (("" ".")) NIL NIL 3 OK Namespace completed.
- LSUB () "." "00000archive" ... 4 OK Lsub completed.
- LIST (\HasNoChildren) "." "INBOX" 5 OK List completed.
- FLAGS (\Answered \Flagged \Deleted \Seen \Draft $MDNSent)
- OK [PERMANENTFLAGS (\Answered \Flagged \Deleted \Seen \Draft $MDNSent \*)] Flags permitted.
- 1153 EXISTS
- 0 RECENT
- OK [UIDVALIDITY 1112486844] UIDs valid
- OK [UIDNEXT 1156] Predicted next UID 6 OK [READ-WRITE] Select completed.
- 1 FETCH (FLAGS (\Seen) UID 1)
- 2 FETCH (FLAGS (\Seen) UID 2)
- 3 FETCH (FLAGS (\Seen) UID 3)
- 4 FETCH (FLAGS (\Seen) UID 4)
- 5 FETCH (FLAGS (\Seen) UID 5) ...
- 1108 FETCH (FLAGS (\Seen) UID 1108)
- 1109 FETCH (FLAGS (\Seen) UID 1109)
- 1110 FETCH (FLAGS (\Seen) UID 1110)
- 1111 FETCH (FLAGS (\Seen) UID 1113)
- 1112 FETCH (FLAGS (\Seen) UID 1114)
- 1113 FETCH (FLAGS (\Seen) UID 1115)
- 1114 FETCH (FLAGS (\Seen) UID 1116)
- 1115 FETCH (FLAGS (\Seen) UID 1117) ...
- 1150 FETCH (FLAGS (\Seen) UID 1152)
- 1151 FETCH (FLAGS (\Seen) UID 1153)
- 1152 FETCH (FLAGS (\Seen) UID 1154)
- 1153 FETCH (FLAGS (\Seen) UID 1155) 7 OK Fetch completed.
- 1 FETCH (UID 1 RFC822.SIZE 3515 FLAGS (\Seen) BODY[HEADER.FIELDS (FROM TO CC SUBJECT DATE MESSAGE-ID PRIORITY X-PRIORITY REFERENCES NEWSGROUPS IN-REPLY-TO CONTENT-TYPE)] {464} ...
- 1153 FETCH (UID 1155 RFC822.SIZE 12977 FLAGS (\Seen) BODY[HEADER.FIELDS (FROM TO CC SUBJECT DATE MESSAGE-ID PRIORITY X-PRIORITY REFERENCES NEWSGROUPS IN-REPLY-TO CONTENT-TYPE)] {395} ... Content-Type: text/plain; charset="UTF-8"
) 8 OK Fetch completed.
but with current cvs:
...
- 1150 FETCH (FLAGS (\Seen) UID 1152)
- 1151 FETCH (FLAGS (\Seen) UID 1153)
- 1152 FETCH (FLAGS (\Seen) UID 1154)
- 1153 FETCH (FLAGS (\Seen) UID 1155) 7 OK Fetch completed. 8 OK Fetch completed.
I got in pretty much everywhere I applied for: CS and biotech in
wow. congratulations!
On Tue, 2006-06-27 at 12:57 +0200, Jakob Hirsch wrote:
but something else: .. 8 UID fetch 1:1110,1113:1155 (UID RFC822.SIZE FLAGS
With this fix too?
2006-06-26 08:35 Timo Sirainen tss@iki.fi
* src/lib-storage/index/index-search.c: Searching (storing,
fetching) with sequence sets containing commas was broken
Timo Sirainen wrote:
So, again, please test the latest CVS or the latest nightly snapshot from http://dovecot.org/nightly/
Quota is working very well now - even with xfs. But I have needed this patch since I first got quota working in january or so. Any reason why this is needed, and is it only on Linux? Several people have reported the same problem... Best regards, Stian --- quota-fs.c.orig 2006-06-18 15:46:20.606456150 +0200 +++ quota-fs.c 2006-06-18 15:43:07.299881841 +0200 @@ -243,9 +243,9 @@ } #endif *value_r = (uint64_t)dqblk.dqb_curblocks * - (uint64_t)root->mount->blk_size / 1024; + (uint64_t)root->mount->blk_size / 1024 / 4096; *limit_r = (uint64_t)dqblk.dqb_bsoftlimit * - (uint64_t)root->mount->blk_size / 1024; + (uint64_t)root->mount->blk_size / 1024 / 4; return 1; }
participants (7)
-
Chris H.
-
Jakob Hirsch
-
Mark Plaksin
-
Reuben Farrelly
-
Stephen Lee
-
Stian Jordet
-
Timo Sirainen