Daniel,

> Should I assume that bugzilla is no longer a useful way to track releases?
> For 3.4.0, there are only 2 resolved bugs, and 6 in a review state.  None
> have been updated in over a month.  The 3.3.2 milestone isn't much better,
> although there are 19 resolved bugs and 88 more targeting that release.
> It also hasn't seen any activity in about a month.

Well, yes, the state as glanced from a bugzilla is not encouraging,
it is true there hasn't been any significant activity for the last few months.

It is important to note that any bug fixes or enhancements scheduled and
done for the 3.3 branch are first committed to trunk (commit-then-review
mode), and only then backported to 3.3 for a review and commit.
So whatever seems good with the 3.3.2 milestone (committed or
just proposed or in a review state) is already in trunk (to become 3.4).

The other important thing to note is that many of the tickets slated
for 3.4.0 (or some still targeted as 'future' or 'undefined') did receive
some solution which is committed to trunk, but they may be waiting
for feedback from a problem reporter or from other members of the
project, or may still need an automated test script or some other
detail finished.

Nevertheless, whatever is commited to trunk did receive testing from
its author, and from Hudson automated testing procedures, and from
people running the trunk code in production or test sites. Any problems
reported in the freshly committed code are sorted out immediately,
usually within hours or a day from a report. So a trunk version that
remains unchanged for a few days it very likely to be stable, with
3.3 bug fixes (that may still be waiting for approval) already fixed, and
with new features in place. I'd choose trunk over 3.3.1 (let alone 3.2.*)
in production any time, at least the way it behaves for the past year.

There may be a problem mixing CPAN perl modules (or SA svn)
with other installed perl modules. It very much depends on your
packageing/installation system and policy. With FreeBSD there are
no problems mixing perl modules from ports with hand-installed
or cpan-installed modules, as the ports system is aware of
hand-installed modules. I've heard that for Linux installations
this is much more of a nightmare, so it is important to follow
the correct procedures there (build a package for a perl module
and install it as a package, but you'd know better).



Below is a list of my commits to SpamAssassin trunk
since the 3.3.1 release (omitting the more trivial entries,
not all committed code is entirely mine).

The more prominent features worth mentioning are:
- the learn-on-error setting;
- new AskDNS plugin, with the DKIMDOMAIN_IN_DWL rule
 for testing against dwl.spamhaus.org;
- Plugin/ASN.pm overhaul;
- whitelist DNS lookups for IPv6 addresses;
- added enlist_uri_host and delist_uri_host conf directives;
- new config directives dns_local_ports_permit,
  dns_local_ports_avoid and dns_local_ports_none;
- instant failover from a dead DNS resolver
- new unwhitelist_from_dkim directive
- fixed warnings with perl 5.12.





------- 2011-01-26 14:41:35 +0100 (Wed, 26 Jan 2011)
Bug 6540: lowercase address in unwhitelist_*, documentation clarification

------- 2011-01-07 20:13:57 +0100 (Fri, 07 Jan 2011)
Bug 6500: clear_originating_ip_headers seems to be broken

------- 2011-01-07 15:30:25 +0100 (Fri, 07 Jan 2011)
Bug 6531: clear_uridnsbl_skip_domain feature to allow admin override
of default configuration

------- 2011-01-06 20:53:03 +0100 (Thu, 06 Jan 2011)
Bug 5655: Bayes not considering add_header'ed information:
let the ASN plugin provide meta-information that can be
tokenized by bayes in a form of X-ASN and X-ASN-Route
header fields

------- 2011-01-06 13:17:01 +0100 (Thu, 06 Jan 2011)
Plugin/ASN.pm overhaul:
- Bug 6484: alternative zone FORMAT enhancements
- recognize response formats using multiple TXT RR string fields
- use new SA::Util::reverse_ip_address() utility
- allow ASN queries for IPv6 addresses
- added config directives clear_asn_lookups and asn_prefix
- use a public API routine $pms->set_tag() for setting
  tag values instead of directly accessing data structure 
- cleanup, some hardening

------- 2010-12-28 15:12:26 +0100 (Tue, 28 Dec 2010)
Bug 6499 and mailing list: wrapped scores for rules DKIMDOMAIN_IN_DWL*,
ACCESSDB and SHORTCIRCUIT into a suitable ifplugin/endif to avoid
lint warnings; removed score for nonexistent rule SUBJ_RE_NUM.

------- 2010-12-15 16:50:50 +0100 (Wed, 15 Dec 2010)
Conf.pm: document tflags interpreted by the URIDNSBL plugin

------- 2010-12-15 03:34:38 +0100 (Wed, 15 Dec 2010)
Bug 6518: Plugin::AskDNS now also accepts DNS rcode as a filtering
subrule, making it possible to distinguish a NXDOMAIN from other
failures; provide tags FIRSTTRUSTEDREVIP and LASTEXTERNALREVIP

------- 2010-12-13 16:27:13 +0100 (Mon, 13 Dec 2010)
Bug 6519, use NetAddr::IP::full6 only when available, without
bumping up the required version

------- 2010-12-01 02:29:01 +0100 (Wed, 01 Dec 2010)
Bug 6519: RBL lookups for IPv6 addresses

------- 2010-12-01 00:31:19 +0100 (Wed, 01 Dec 2010)
Bug 6460: RCVD_ILLEGAL_IP false positives

------- 2010-11-30 19:24:40 +0100 (Tue, 30 Nov 2010)
Bug 6499: Add rules for new Spamhaus whitelisting DWL

------- 2010-11-30 19:10:11 +0100 (Tue, 30 Nov 2010)
Bug 6518: Implement a general purpose Plugin::AskDNS, makes
Spamhaus DWL possible

------- 2010-11-30 17:35:15 +0100 (Tue, 30 Nov 2010)
Bug 6517: New infrastructure triggering actions when their
dependency tags become available

------- 2010-11-04 18:32:04 +0100 (Thu, 04 Nov 2010)
Bug 6508: Speeding up lookups on {trusted,internal,msa}_networks

------- 2010-11-04 01:59:40 +0100 (Thu, 04 Nov 2010)
Bug 6506: Modifying a list while traversing it with a foreach

------- 2010-10-28 02:39:18 +0200 (Thu, 28 Oct 2010)
Bug 6488 - Lint errors with Perl 5.12.1 in AntiVirus.pm

------- 2010-10-27 18:54:31 +0200 (Wed, 27 Oct 2010)
Bug 6380: sa-update should handle case where some channel(s)
fail but others work - introduce exit status 3

------- 2010-10-25 17:15:42 +0200 (Mon, 25 Oct 2010)
Bug 6392: fix one more case of a 'goto into a construct',
this one occurred with sa-compile

------- 2010-09-02 10:42:53 +0200 (Thu, 02 Sep 2010)
20_freemail_domains.cf: runbox.com is not a free email provider since 2001

------- 2010-07-26 21:08:16 +0200 (Mon, 26 Jul 2010)
Bug 6468: splice() offset past end of array in HTML.pm

------- 2010-06-29 01:29:02 +0200 (Tue, 29 Jun 2010)
Bug 6458 - add enlist_uri_host and delist_uri_host conf directives,
allowing for arbitrarily named URI lists, each associated with
its own scoring rule

------- 2010-06-24 18:54:54 +0200 (Thu, 24 Jun 2010)
Bug 6458: fix handling user_prefs for uri wblisting

------- 2010-06-24 03:50:56 +0200 (Thu, 24 Jun 2010)
Bug 6458: add blacklist_uri_host, whitelist_uri_host;
and A record lookups to URIs in URIDNSBL plugin

------- 2010-06-20 21:44:48 +0200 (Sun, 20 Jun 2010)
some small optimizations to URIDNSBL.pm: factor out dereferencing
$scanner->{conf} from loops;  some refs to {conf} were through
$scanner->{main}->{conf} while others through $scanner->{conf},
standardize to the later for consistency;  remove vestiges of
log_dns_result() and call dbg directly in remaining few places
(same overhead: one routine call);  don't count number of keys
of a hash when all we need is to know if it is empty or not

------- 2010-06-16 18:58:20 +0200 (Wed, 16 Jun 2010)
Bug 6406: FROM_STARTS_WITH_NUMS matches on text-to-email

------- 2010-06-16 01:42:46 +0200 (Wed, 16 Jun 2010)
Bug 6454: wrong status test on $sth->rows in BayesStore::PgSQL

------- 2010-06-16 01:05:01 +0200 (Wed, 16 Jun 2010)
tweaking timing instrumentation on bayes (bug 6444, bug 4400)

------- 2010-06-14 20:26:40 +0200 (Mon, 14 Jun 2010)
related to bug 4400: instrument the Bayes pluging
with more fine-grained timing report, to also show
processing times in tokenization and in computing
probabilities

------- 2010-06-10 19:08:24 +0200 (Thu, 10 Jun 2010)
Bug 6444: tok_touch_all update forces full table scan, kills performance.

------- 2010-06-09 19:06:25 +0200 (Wed, 09 Jun 2010)
Bug 6447: new feature to bayes autolearning: learn-on-error

------- 2010-06-08 17:57:37 +0200 (Tue, 08 Jun 2010)
Bug 6418: Cannot Log to stderr without timestamps;
provide option --log-timestamp-fmt to spamd

------- 2010-06-04 15:19:50 +0200 (Fri, 04 Jun 2010)
Bug 6377: win32: spamd signal handling

------- 2010-06-03 20:41:26 +0200 (Thu, 03 Jun 2010)
Bug 6443: Metadata header field names should not be case-sensitive,
for consistency with regular (in-message) header field names.

------- 2010-06-03 18:48:30 +0200 (Thu, 03 Jun 2010)
Bug 5491: MIME_QP_LONG_LINE triggering on valid email:
change a QP line limit from 76 to 78

------- 2010-06-01 21:21:38 +0200 (Tue, 01 Jun 2010)
avoid undef warning: Use of uninitialized value $opt{"syslog-socket"}
in lc at spamd line 444

------- 2010-06-01 15:51:01 +0200 (Tue, 01 Jun 2010)
Bug 6376: win32: consider negative pids under windows in spamd's waitpid

------- 2010-05-31 20:46:11 +0200 (Mon, 31 May 2010)
Bug 6382 - add missing unwhitelist_from_dkim, part deux

------- 2010-05-31 17:44:52 +0200 (Mon, 31 May 2010)
Bug 6375: posix macro not implemented in win32 - spamd

------- 2010-05-30 15:00:47 +0200 (Sun, 30 May 2010)
Bug 5894: spamd doesn't use vpopmail virtual users' dirs - removed
one extra space

------- 2010-05-30 14:50:52 +0200 (Sun, 30 May 2010)
Bug 6325: FH_HAS_XID Poor S/O, strong score - zeroing it out

------- 2010-05-13 20:07:04 +0200 (Thu, 13 May 2010)
Bug 6382#6: removing serv...@youtube.com and amazon from
def_whitelist_from_dkim

------- 2010-04-22 03:44:59 +0200 (Thu, 22 Apr 2010)
Bug 6404: Fix perldoc errors

------- 2010-04-19 17:22:46 +0200 (Mon, 19 Apr 2010)
Bug 6408: fix more cases of case-sensitive URI schemes

------- 2010-04-18 23:51:57 +0200 (Sun, 18 Apr 2010)
Untaint a list of modules given by an auto_whitelist_db_modules
configuration directive, avoiding a taint failure in DBBasedAddrList

------- 2010-04-07 18:16:53 +0200 (Wed, 07 Apr 2010)
avoid undef warnings in AutoWhitelist.pm as a result of
incorrect Received header field or its incorrect parsing
(as in Bug 6402)

------- 2010-03-31 02:02:30 +0200 (Wed, 31 Mar 2010)
Bug 6396: Use of uninitialized value in lc
at lib/Mail/SpamAssassin/Plugin/MIMEEval.pm

------- 2010-03-30 18:03:42 +0200 (Tue, 30 Mar 2010)
Bug 6392: Test suite fails with perl 5.12.0:
avoid goto run_compiled_method in Plugin::Check, getting rid
of 'Use of "goto" to jump into a construct is deprecated'

------- 2010-03-30 16:20:03 +0200 (Tue, 30 Mar 2010)
Bug 6221: DnsResolver.pm attempts to bind ports at random:
make auxiliary data structures initializations more
self-contained to avoid undef warnings in t/dnsbl_subtests.t;
save about 10 kB when port allocation task is delegated to OS

------- 2010-03-30 13:20:58 +0200 (Tue, 30 Mar 2010)
Bug 6392: Test suite fails with perl 5.12.0;
some more cleanups according to Bug 6392 comment 4

------- 2010-03-30 03:33:52 +0200 (Tue, 30 Mar 2010)
Stay on the safe side regarding taintedness of $1,$2,$3,$4.

------- 2010-03-30 03:29:47 +0200 (Tue, 30 Mar 2010)
defined(%hash) is not the way to test a hash for being empty

------- 2010-03-30 03:18:22 +0200 (Tue, 30 Mar 2010)
Bug 4260: rewrite DNS code to use a single socket, event-based model;
Bug 6338: Use of Bit 0x20 in DNS Labels to Improve Transaction Identity;
Remove case sensitivity on DNS label in a query section
when 'dns_options dns0x20' is NOT used - RFC 1035 does NOT require
DNS server to preserve case in a query section of a reply, even
though in practice all mainstream DNS servers do preserve it.
In effect, code before this patch was half-way between dns0x20
and nodns0x20 setting.

------- 2010-03-29 19:40:40 +0200 (Mon, 29 Mar 2010)
Bug 5264: Specification of DNS-query source port range;
Bug 6221: DnsResolver.pm attempts to bind ports at random;
Implements config directives dns_local_ports_permit,
dns_local_ports_avoid and dns_local_ports_none, allowing
admin control over the set of local port numbers used
in sending DNS queries to a resolving DNS server, or
letting the operating system choose a free port number.
Also avoids a sequential ports scanning when one is
unlucky that a random start point falls in a range
of consecutive busy ports.
The 10_default_prefs.cf now brings a set of
dns_local_ports_avoid directives to exclude IANA
assigned high port numbers and some other ranges.

------- 2010-03-25 13:41:26 +0100 (Thu, 25 Mar 2010)
mention perl #67962 bugfix in a comment

------- 2010-03-24 01:41:05 +0100 (Wed, 24 Mar 2010)
Bug 6385: Instant failover from a dead DNS resolver

------- 2010-03-19 18:53:34 +0100 (Fri, 19 Mar 2010)
Bug 6382: removed facebook and linkedin from def_whitelist_from_dkim

------- 2010-03-19 18:48:27 +0100 (Fri, 19 Mar 2010)
Bug 6382: add in missing unwhitelist_from_dkim


  Mark

------------------------------------------------------------------------------
The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
http://p.sf.net/sfu/intel-dev2devfeb
_______________________________________________
AMaViS-user mailing list
AMaViS-user@lists.sourceforge.net 
https://lists.sourceforge.net/lists/listinfo/amavis-user 
 Please visit http://www.ijs.si/software/amavisd/ regularly
 For administrativa requests please send email to rainer at openantivirus dot 
org

Reply via email to