This is the Changes file with a large number of entries that seemed less important removed to make it easier to generate the Summary of Changes for the release announcement. I considered anything to do with mass checks, rule-qa and other development tools to not be needed for the Summary Note that I only did a quick scan, leaving many entries that are not important enough to go into the summary of changes for the release announcement. But this should be easier to work with than the full Changes file. You probably need to say something that covers the aggregate of many of the entries that I removed such as "Many minor bugs fixed, rules that are no longer effective removed, new rules added, and code optimised for speed and memory". Also, something about sa-compile and changes to sa-update -- sidney ------------------------------------------------------------------------ r520881 | jm | 2007-03-22 01:30:05 +1200 (Thu, 22 Mar 2007) | 1 line bug 5365: spamd child falling back to 'nobody' due to unknown username passed from spamc, attempts to read previous user's bayes dbs (and fails). fixed ------------------------------------------------------------------------ r518982 | jm | 2007-03-16 14:09:14 +0000 (Fri, 16 Mar 2007) | 1 line bug 5338: check Config{installsiteman1dir} and Config{installvendorman1dir} as well as Config{installman1dir}, otherwise we can wind up not installing many of the important man pages when we should do so ------------------------------------------------------------------------ r518664 | dos | 2007-03-15 15:34:08 +0000 (Thu, 15 Mar 2007) | 1 line prevent plugins from permanently disabling their dns tests, dns may become available again with re-testing of dns availability ------------------------------------------------------------------------ r518631 | jm | 2007-03-15 14:05:34 +0000 (Thu, 15 Mar 2007) | 1 line bug 5344: binmode() causes 'CR NL' ('\r\n') to be read as 'NL' ('\n'). fix this by using the octal escape sequences: \015\012, instead, in the core spots where we access user-submitted mail data through a filehandle that may be affected by binmode. ------------------------------------------------------------------------ r518443 | dos | 2007-03-15 02:21:34 +0000 (Thu, 15 Mar 2007) | 1 line add clear_msa_networks to tests where needed ------------------------------------------------------------------------ r518187 | jm | 2007-03-14 15:38:32 +0000 (Wed, 14 Mar 2007) | 1 line refactor some shared code from spamd_hup.t, spamd_kill_restart.t and spamd_kill_restart_rr.t into SATest.pm; add new more-raceproof spamd_hup logic to other two tests, too ------------------------------------------------------------------------ r518167 | jm | 2007-03-14 15:00:34 +0000 (Wed, 14 Mar 2007) | 1 line add some more sanity checks on pid file size ------------------------------------------------------------------------ r518147 | jm | 2007-03-14 14:00:03 +0000 (Wed, 14 Mar 2007) | 1 line another attempt at a race-resistant spamd SIGHUP test case ------------------------------------------------------------------------ r518117 | jm | 2007-03-14 12:40:07 +0000 (Wed, 14 Mar 2007) | 1 line add alternative fitness function: distance from a target false-positive rate. This is easier to predict, and produces better GA results in fewer generations in 10-pass cross-validation tests. However, it's not actually active yet in runGA; next time the GA is run, it will be worthwhile switching to this ------------------------------------------------------------------------ r518111 | jm | 2007-03-14 12:31:49 +0000 (Wed, 14 Mar 2007) | 1 line GA improvements. remove long-vestigial USE_LOG_SCORE_EVALUATION and exhaustive_eval code; make maxiter a command-line parameter; add a digit to the percentage counters in the output; silence some inscrutable noise in the output; and make the checkpoints output more frequently ------------------------------------------------------------------------ r517750 | jm | 2007-03-13 16:09:42 +0000 (Tue, 13 Mar 2007) | 1 line this should be an info, so people can know the compiled rules are in use [Editor's note: sa-compile, compiled rules, implemented. I'm deleting all following checkin notices for them] ------------------------------------------------------------------------ r516154 | jm | 2007-03-08 19:19:31 +0000 (Thu, 08 Mar 2007) | 1 line bug 5352: use zero-space lookbehind assertion to avoid ACM hosts ------------------------------------------------------------------------ r515725 | dos | 2007-03-07 20:00:43 +0000 (Wed, 07 Mar 2007) | 1 line bug 5348: add 25_asn.cf for ASN plugin and cleanup/clairify plugin POD ------------------------------------------------------------------------ r515540 | jm | 2007-03-07 11:42:03 +0000 (Wed, 07 Mar 2007) | 1 line add an UPGRADE section for 3.2.0, listing the 127/8-trusted issue and the new msa_networks setting; more to follow, no doubt ------------------------------------------------------------------------ r515298 | jm | 2007-03-06 21:00:54 +0000 (Tue, 06 Mar 2007) | 43 lines A couple of enhancements to Plugin::DKIM from Mark Martinec: 'prompted by recent enhancements to the underlying module Mail::DKIM, which now also supports DomainKeys signatures in addition to DKIM signatures, and is being actively maintained, unlike Mail::DomainKeys. - separate a signature verification from fetching a policy, which makes it possible to avoid one DNS lookups (by not fetching a policy) for each unverified message by setting score to 0 for all policy-related rules (DKIM_POLICY_SIGNALL, DKIM_POLICY_SIGNSOME, and DKIM_POLICY_TESTING). - let the check_dkim_testing() also take into account a testing flag in a public key, not just the one in a policy record (still lacks the final support in Mail::DKIM (missing a documented method), but the plugin is now ready). This is an equivalent change that was already made to Plugin::DomainKeys. - skip fetching a policy (SSP) if a signature does verify, according to draft-allman-dkim-ssp-02: If the message contains a valid Originator Signature, no Sender Signing Practices check need be performed: the Verifier SHOULD NOT look up the Sender Signing Practices and the message SHOULD be considered non-Suspicious. This is an equivalent change that was already made to Plugin::DomainKeys. - make some debugging messages more informative or concise; - bug fix: protect fetching/parsing a policy record by eval { } to prevent a syntax error in a policy record from invalidating an already verified good signature; - bug fix: if an identity tag (i=) is missing (this tag is optional), it should default to an '@' prepended to a domain (d=), according to draft-ietf-dkim-base-09; previously dkim whitelisting check was skipped when identity was not explicit ('i' tag missing); - squash code in _check_dkim_whitelist() in half by factoring out duplicate code sections ------------------------------------------------------------------------ r515222 | jm | 2007-03-06 17:36:12 +0000 (Tue, 06 Mar 2007) | 1 line make it clear that Mail::DKIM is better than Mail::DomainKeys ------------------------------------------------------------------------ r513691 | jm | 2007-03-02 10:21:49 +0000 (Fri, 02 Mar 2007) | 1 line bug 5362: add Mail::SPF as an optional module, and document its relationship to Mail::SPF::Query ------------------------------------------------------------------------ r513418 | jm | 2007-03-01 17:45:29 +0000 (Thu, 01 Mar 2007) | 1 line bug 5360: spamd --daemonize fails to start if the MySQL or SQL BayesStore backends are in use, due to SIGPIPE delivered before signal handlers are set up. fix ------------------------------------------------------------------------ r512262 | jm | 2007-02-27 14:14:16 +0000 (Tue, 27 Feb 2007) | 1 line more portability-friendly makefile; LIBDIR has to be update between solaris and linux. also, fix deps for garescorer binary ------------------------------------------------------------------------ r511876 | felicity | 2007-02-26 16:22:53 +0000 (Mon, 26 Feb 2007) | 1 line bug 5353: don't let boolean not match when neq is supposed to ... ------------------------------------------------------------------------ r511346 | jm | 2007-02-24 21:09:00 +0000 (Sat, 24 Feb 2007) | 1 line bug 5257: set autolearn thresholds to be more conservative for learning mails as ham, due to anecdotal reports of problems ------------------------------------------------------------------------ r511345 | jm | 2007-02-24 21:08:38 +0000 (Sat, 24 Feb 2007) | 1 line bug 4747: update documentation to match current EnvelopeFrom behaviour ------------------------------------------------------------------------ r508709 | felicity | 2007-02-17 05:01:42 +0000 (Sat, 17 Feb 2007) | 1 line bug 5337: change the start order for redhat such that spamd starts before the MTA ------------------------------------------------------------------------ r508375 | jm | 2007-02-16 12:01:00 +0000 (Fri, 16 Feb 2007) | 1 line [bug 5318: set a maximum internal length for URIs] increase URI length limit to 8k as discussed in bug 5318 comment 30 ------------------------------------------------------------------------ r508076 | jm | 2007-02-15 18:59:01 +0000 (Thu, 15 Feb 2007) | 1 line bug 5179: perl does silly things with \r\n line endings on non-UNIX platforms. use \015\012 instead ------------------------------------------------------------------------ r507897 | jm | 2007-02-15 11:36:38 +0000 (Thu, 15 Feb 2007) | 1 line reinstate GA evolver code from r6318, due to issues with perceptron ------------------------------------------------------------------------ r507521 | jm | 2007-02-14 13:06:37 +0000 (Wed, 14 Feb 2007) | 1 line fix reported problem with VBounce plugin, caused by bug 5331; certain MIME structures are unparseable using Message::find_parts(), so just use the slower but more reliable full message rendering instead ------------------------------------------------------------------------ r507505 | jm | 2007-02-14 12:16:11 +0000 (Wed, 14 Feb 2007) | 1 line bug 5328: always use an absolute path for the apache-spamd PidFile setting, thanks to Radoslaw Zielinski ------------------------------------------------------------------------ r507503 | jm | 2007-02-14 12:14:52 +0000 (Wed, 14 Feb 2007) | 1 line bug 4329: support --cf 'config line' option in apache-spamd, thanks to Radoslaw Zielinski ------------------------------------------------------------------------ r504908 | jm | 2007-02-08 14:27:55 +0000 (Thu, 08 Feb 2007) | 1 line bug 5285: reuse NJABL_DUL Dynablock hits as input for RCVD_IN_PBL during the perceptron run ------------------------------------------------------------------------ r504893 | jm | 2007-02-08 13:46:22 +0000 (Thu, 08 Feb 2007) | 1 line bug 5271: more ImageInfo integration; update CREDITS to note that it was contributed by Dallas E. Engelken (thanks Dallas); add loadplugin line to MANIFEST; and comment out the rules that didn't reach quality levels in the ruleqa system (were still sandbox rules) ------------------------------------------------------------------------ r504888 | jm | 2007-02-08 13:39:33 +0000 (Thu, 08 Feb 2007) | 1 line bug 4271: move ImageInfo into 3.2.0 core ruleset ------------------------------------------------------------------------ r504605 | felicity | 2007-02-07 16:18:09 +0000 (Wed, 07 Feb 2007) | 1 line bug 5240: also disable auto_whitelist_factory in updates ------------------------------------------------------------------------ r503963 | felicity | 2007-02-06 02:45:08 +0000 (Tue, 06 Feb 2007) | 1 line bug 5240: also disable bayes_store_module without --allowplugins ------------------------------------------------------------------------ r503495 | jm | 2007-02-04 21:59:33 +0000 (Sun, 04 Feb 2007) | 1 line bug 4576: handle mail.gmx.net-style qmail headers ------------------------------------------------------------------------ r502116 | felicity | 2007-02-01 04:15:39 +0000 (Thu, 01 Feb 2007) | 1 line bug 5261: only try to match terminal size w/ regexp if variable is defined ------------------------------------------------------------------------ r501206 | jm | 2007-01-29 22:32:31 +0000 (Mon, 29 Jan 2007) | 1 line bug 5311: die() if no plugin implements 'check_main' and the check() API is called, otherwise it's reasonably easy to wind up with a non-scanning scanner. also, add a test ------------------------------------------------------------------------ r500235 | jm | 2007-01-26 13:20:01 +0000 (Fri, 26 Jan 2007) | 1 line add new Devel::SawAmpersand-based test, which detects use of the dollar-ampersand perlvar -- this slows down every regexp match performed in the perl interpreter afterwards, so is better avoided in SpamAssassin. test is SVN-only and requires editing t/config to run ------------------------------------------------------------------------ r500052 | dos | 2007-01-25 23:26:59 +0000 (Thu, 25 Jan 2007) | 1 line bug 5305: add msa_networks to the list of config items with their own clone method ------------------------------------------------------------------------ r500015 | felicity | 2007-01-25 22:12:53 +0000 (Thu, 25 Jan 2007) | 3 lines Add in my samailoffset script which can be used to retrieve messages from a corpus, mostly for use with mass-check logs. ------------------------------------------------------------------------ r499778 | jm | 2007-01-25 13:03:52 +0000 (Thu, 25 Jan 2007) | 1 line bug 5304: move documentation for whitelist_from, whitelist_from_rcvd, unwhitelist_from, def_whitelist_from_rcvd, whitelist_allows_relays, unwhitelist_from_rcvd, blacklist_from, unblacklist_from, whitelist_to, more_spam_to, all_spam_to, and blacklist_to, back to Conf.pm from Mail/SpamAssassin/Plugin/WLBLEval.pm -- since they are basic config settings, frequently used by endusers and documentation needs to be easily accessible. ------------------------------------------------------------------------ r499774 | jm | 2007-01-25 12:58:08 +0000 (Thu, 25 Jan 2007) | 1 line bug 5295: add 'whitelist_auth', to whitelist addresses that send mail using sender-authorization systems like SPF, Domain Keys, and DKIM ------------------------------------------------------------------------ r499691 | dos | 2007-01-25 07:22:38 +0000 (Thu, 25 Jan 2007) | 9 lines - fix weak validation regexes in netset that caused "trusted_networks 10/8 !10/8" to be accepted as valid input (only complete dot-quads were being checked for duplication) - redirect STDERR in t/trust_path.t when expecting netset to warn about bad trusted/internal/msa networks config; also cause the tests to fail if the expected warnings aren't triggered ------------------------------------------------------------------------ r499613 | dos | 2007-01-24 23:25:25 +0000 (Wed, 24 Jan 2007) | 36 lines bug 5305: implement msa_networks for detecting MSAs and extending trust accordingly If a host acts solely as an MSA (and not a combined MX/MSA and/or intermediate relay) we can rely on it to take care of making sure its clients pass some sort of auth check, and thus are trusted (trusted in the sense of SAs trusted_networks). Similar functionality exists now so long as the MSA included an auth token in its Received header which is supported by SA. msa_networks supports identifying clients of MSAs that don't include any sort of visibile auth token, such as older versions of Postfix and whatever it is that gmx.net is using (probably old Postfix). Supporting MSAs without visible auth tokens (which includes new versions of Postfix which has visible auth tokens disabled by default) also satisfies the vocal minority who are super paranoid (for some unknown reason) about others knowing that a client passed some sort of auth check before relaying mail. Usage notes: If your MSA acts solely as an MSA, put the MSA in all three of trusted_networks, internal_networks, msa_networks. If your MSA is also acting as an MX server never, NEVER, include it in msa_networks since you'll end up trusting everyone who submits mail to the MX server. If your MSA is also acting as an intermidiate relay that may receive mail that was submitted to an MX server never, NEVER, include it in msa_networks since you'll end up trusting everyone who submits mail to the MX server. ------------------------------------------------------------------------ r499388 | jm | 2007-01-24 13:00:50 +0000 (Wed, 24 Jan 2007) | 1 line bug 5268: support single-digit hexadecimal numbers in the host part of URIs ------------------------------------------------------------------------ r499012 | jm | 2007-01-23 13:08:29 +0000 (Tue, 23 Jan 2007) | 1 line remove_spamassassin_markup() / spamassassin -d was broken with CR-LF line endings; it replaced them with LF. fix, and add test case to strip2.t to prevent future regression. Also, add a test case for markup (and removal) on a message with no body ------------------------------------------------------------------------ r498825 | jm | 2007-01-22 22:05:21 +0000 (Mon, 22 Jan 2007) | 1 line bug 5250: in rewrite_mail(), messages would have all line endings rewritten -- even inside 8bit binary parts. fix ------------------------------------------------------------------------ r498650 | jm | 2007-01-22 15:29:12 +0000 (Mon, 22 Jan 2007) | 1 line bug 5300: DKIM signature verification would fail to verify if a message contained DOS-style line endings. fix from Jason Long ------------------------------------------------------------------------ r498595 | jm | 2007-01-22 12:31:32 +0000 (Mon, 22 Jan 2007) | 1 line bug 4770: re-apply Mail::SpamAssassin::Plugin::ASN patch, now that licensing is sorted. exposes ASN data as a Bayes token and the _ASNCIDR_ and _ASN_ header-rewriting tags. thanks to Matthias Leisi ------------------------------------------------------------------------ r497460 | jm | 2007-01-18 15:38:38 +0000 (Thu, 18 Jan 2007) | 1 line bug 5172: more woes with stupid EAI_FOO error codes from getaddrinfo(); it turns out that not only does Cygwin omit some of them, so does FreeBSD 6.1p10. Work around by using the magic of autoconf for all of: EAI_ADDRFAMILY, EAI_SYSTEM, EAI_NODATA. ------------------------------------------------------------------------ r497066 | jm | 2007-01-17 16:29:07 +0000 (Wed, 17 Jan 2007) | 1 line oops, more RCVD_IN_NJABL_DUL work; remove descriptions from i18n files, and delete its scores. Instead, copy its scores for RCVD_IN_PBL ------------------------------------------------------------------------ r497041 | jm | 2007-01-17 15:09:18 +0000 (Wed, 17 Jan 2007) | 1 line bug 4728: fix -notfirsthop DNSBL lookup rules to use -lastexternal instead, since it reduces FPs and is easier for legit senders to avoid ------------------------------------------------------------------------ r497038 | jm | 2007-01-17 15:05:23 +0000 (Wed, 17 Jan 2007) | 1 line bug 5187: move RCVD_IN_PBL to main ruleset now that it's been released; bug 5285: retire RCVD_IN_NJABL_DUL in favour of RCVD_IN_PBL ------------------------------------------------------------------------ r497030 | jm | 2007-01-17 14:52:17 +0000 (Wed, 17 Jan 2007) | 1 line bug 5296: add spamc --headers switch, which scans messages and transmits back just rewritten headers. This is more bandwidth-efficient than the normal mode of scanning, but only works for 'report_safe 0'. Bump spamc/spamd's protocol version to 1.4, to reflect new HEADERS verb. update spamd/PROTOCOL for current protocol. add 'sa-compile' to the SVN ignored-files list. ------------------------------------------------------------------------ r496086 | jm | 2007-01-14 14:57:11 +0000 (Sun, 14 Jan 2007) | 1 line bug 4999: ensure spamd --auth-ident requires ident match User header in all situations ------------------------------------------------------------------------ r496085 | jm | 2007-01-14 14:36:05 +0000 (Sun, 14 Jan 2007) | 1 line bug 5294: RCVD_IN_XBL should use -lastexternal, not -notfirsthop ------------------------------------------------------------------------ r495649 | jm | 2007-01-12 17:21:25 +0000 (Fri, 12 Jan 2007) | 1 line bug 5283: Message::get_pristine_header() should produce tainted output, since it's a section of the raw, tainted, input message. Add taint_var() API to Util.pm to support this. also, t/tainted_msg.t was being skipped on too many platforms unnecessarily ------------------------------------------------------------------------ r495220 | jm | 2007-01-11 13:50:42 +0000 (Thu, 11 Jan 2007) | 1 line bug 5278: remove 6-month limit imposed via the __UNUSABLE_MSGID rule on FORGED_MUA_* rules which use Message-ID header ------------------------------------------------------------------------ r494934 | felicity | 2007-01-10 19:06:47 +0000 (Wed, 10 Jan 2007) | 1 line bug 5289: if ignoring a message during learning, set the return code such that we didn't learn from it, not that there was an error during learning which could cause the caller to abort, etc. ------------------------------------------------------------------------ r494481 | jm | 2007-01-09 16:46:01 +0000 (Tue, 09 Jan 2007) | 1 line bug 4471: add --connect-retries and --retry-sleep switches to spamc, thanks to John Madden ------------------------------------------------------------------------ r494446 | jm | 2007-01-09 16:05:08 +0000 (Tue, 09 Jan 2007) | 1 line bug 5286: fix spamc build failure when zlib is not available, fix from John Madden ------------------------------------------------------------------------ r494435 | jm | 2007-01-09 15:32:38 +0000 (Tue, 09 Jan 2007) | 1 line bug 5112: the default Debian configuration for Exim does not record a HELO string if the HELO matched reverse DNS; take this into account ------------------------------------------------------------------------ r494370 | jm | 2007-01-09 10:52:00 +0000 (Tue, 09 Jan 2007) | 1 line bug 4687: exempt messages in 'big5' charset from MIME_BASE64_TEXT rule ------------------------------------------------------------------------ r494093 | jm | 2007-01-08 15:33:53 +0000 (Mon, 08 Jan 2007) | 1 line add spamc '-z' switch, which compresses mails to be scanned using zlib compression; very useful for long-distance use of spamc over the internet. also add test script, INSTALL doc, and documentation. backport from r481882 on 'jm_spamc_hacks' branch ------------------------------------------------------------------------ r493708 | jm | 2007-01-07 12:38:10 +0000 (Sun, 07 Jan 2007) | 1 line bug 5282: document that PERL5LIB is ignored, due to a feature in taint checking ------------------------------------------------------------------------ r493217 | dos | 2007-01-05 22:41:27 +0000 (Fri, 05 Jan 2007) | 29 lines bug 5259: always trust 127/8 no matter how trusted/internal networks are configured, or even if they aren't configured no provision for removing 127/8 from trusted/internal networks is provided since there is no reason why they cannot be trusted since you must trust the machine SA is running on (which could add local hops) and you'd have to first trust another relay that has it's own local hops before local hops on that other machine were considered trusted/internal if you think there's a reason to not consider 127/8 trusted/internal (even if it's mail submitted from a web server on the same machine) you're almost certainly misunderstading the meaning of trusted/internal networks bug 5235: when trusted_networks are inferred (not manually configured) infer the same settings for internal_networks as we would for trusted_networks this makes the inferral method consistent with how we use the same settings for both trusted and internal networks when only one of them are set manually this resolves a number of FP issues with tests, such as SPF and a number of DNSBL tests, that need to know where the handoff is between the remote and local domains ------------------------------------------------------------------------ r493029 | jm | 2007-01-05 13:26:39 +0000 (Fri, 05 Jan 2007) | 1 line bug 5273: add _LASTEXTERNALIP_, _LASTEXTERNALRDNS_, _LASTEXTERNALHELO_ markup tags, to allow adding the connecting IP/HELO as markup to message headers ------------------------------------------------------------------------ r492719 | felicity | 2007-01-04 20:24:32 +0000 (Thu, 04 Jan 2007) | 1 line bug 5051: deal with empty message/* parts appropriately ------------------------------------------------------------------------ r492202 | jm | 2007-01-03 16:38:50 +0000 (Wed, 03 Jan 2007) | 1 line bug 4820: fix get_and_create_userstate_dir() to create the .spamassassin directory in the correct location, if virtual users are being used (thanks to ) ------------------------------------------------------------------------ r491858 | jm | 2007-01-02 17:26:34 +0000 (Tue, 02 Jan 2007) | 1 line bug 5629: scope v3.2.0 updates with a 'magic' string so they won't collide with 3.1.x updates (or any future 3.3.x, 3.4.x, etc line) ------------------------------------------------------------------------ r491726 | sidney | 2007-01-02 00:27:20 +0000 (Tue, 02 Jan 2007) | 1 line bug 5254: allow for syswrite doing partial writes, fixing a problem with using SSL on some platforms ------------------------------------------------------------------------ r491718 | sidney | 2007-01-02 00:16:58 +0000 (Tue, 02 Jan 2007) | 1 line bug 5266: patch to treat command line argument space in a read-only fashion ------------------------------------------------------------------------ r491592 | sidney | 2007-01-01 15:41:34 +0000 (Mon, 01 Jan 2007) | 1 line bug 5265: Add SSL protocol options to spamc and spamd and make the default be a better choice for working on various platforms ------------------------------------------------------------------------ r491585 | sidney | 2007-01-01 15:18:15 +0000 (Mon, 01 Jan 2007) | 1 line bug 5263: fix bugs when a longopt with no arguments is last on the command line ------------------------------------------------------------------------ r491367 | felicity | 2006-12-31 02:29:53 +0000 (Sun, 31 Dec 2006) | 1 line bug 5009: add in whitespace for both opening and closing whitespace tags ------------------------------------------------------------------------ r491327 | felicity | 2006-12-30 21:50:28 +0000 (Sat, 30 Dec 2006) | 1 line bug 5249: mail could have 8-bit data, need to have ArchiveIterator do binmode on some platforms ------------------------------------------------------------------------ r491312 | felicity | 2006-12-30 21:11:22 +0000 (Sat, 30 Dec 2006) | 1 line bug 4747: use EnvelopeFrom:addr when we want the address and not the full header ------------------------------------------------------------------------ r491304 | felicity | 2006-12-30 20:21:18 +0000 (Sat, 30 Dec 2006) | 1 line bug 5256: if we short-circuit, don't do auto-learning ------------------------------------------------------------------------ r491203 | felicity | 2006-12-30 10:00:08 +0000 (Sat, 30 Dec 2006) | 1 line optimize the bayes stop-list via Regexp::Trie, add a pointer to the wiki page which has the word list, etc. ------------------------------------------------------------------------ r490760 | jm | 2006-12-28 16:44:58 +0000 (Thu, 28 Dec 2006) | 1 line bug 5255: SIGHUP handler couldn't restart spamd if it had been started using an odd commandline (possibly only on solaris) ------------------------------------------------------------------------ r490264 | felicity | 2006-12-26 05:30:50 +0000 (Tue, 26 Dec 2006) | 1 line bug 4459: update RCVD_ILLEGAL_IP to catch more illegal IPs, and allow 127.0.0.* since they appear in ham a bit ------------------------------------------------------------------------ r490114 | felicity | 2006-12-25 07:47:58 +0000 (Mon, 25 Dec 2006) | 1 line bug 3466: do the bayes expiry after results have been passed back to the client from spamd, helps avoid client timeouts, etc. ------------------------------------------------------------------------ r489341 | jm | 2006-12-21 12:45:35 +0000 (Thu, 21 Dec 2006) | 1 line bug 5240: add sa-update --allowplugins switch, and automatically comment all 'loadplugin' or 'tryplugin' lines in downloaded rules files unless it is specified ------------------------------------------------------------------------ r489084 | jm | 2006-12-20 13:29:18 +0000 (Wed, 20 Dec 2006) | 1 line when using format=detect, format would be set by the first file/dir/mbox in a glob and then reused by all other locations; this was breaking the bb- mass-checks ------------------------------------------------------------------------ r488376 | jm | 2006-12-18 18:02:29 +0000 (Mon, 18 Dec 2006) | 1 line bug 4882: don't use relative paths in @INC in the installed SpamAssassin scripts; just in the uncompiled build tree ------------------------------------------------------------------------ r487940 | dos | 2006-12-17 04:44:30 +0000 (Sun, 17 Dec 2006) | 1 line bug 5239: we can run SPF checks in local_tests_only mode now, re-using Received-SPF header results ------------------------------------------------------------------------ r487934 | dos | 2006-12-17 03:17:07 +0000 (Sun, 17 Dec 2006) | 1 line bug 5239: move hackish M::SA::M:Node crap to M::SA::PMS where it should be ------------------------------------------------------------------------ r487789 | jm | 2006-12-16 11:54:04 +0000 (Sat, 16 Dec 2006) | 1 line hack hack hack; update P595Body plugin to use the new blead feature of REGMARK, thanks to demerphq. still needs speeding up unfortunately ------------------------------------------------------------------------ r487661 | felicity | 2006-12-15 21:09:04 +0000 (Fri, 15 Dec 2006) | 1 line bug 5246: drop INFO_TLD due to high FP rate ------------------------------------------------------------------------ r487586 | jm | 2006-12-15 16:03:23 +0000 (Fri, 15 Dec 2006) | 1 line bug 5243: add Plugin::register_method_priority() API, allowing plugins to control the relative ordering of plugin callbacks relative to other plugins' implementations ------------------------------------------------------------------------ r487146 | dos | 2006-12-14 09:21:45 +0000 (Thu, 14 Dec 2006) | 31 lines bug 5239: use results from Received-SPF header where possible - implements the ability to get a range of headers from get_all_headers() using the received headers as index points - adds "ALL-TRUSTED", "ALL-INTERNAL", "ALL-UNTRUSTED" and "ALL-EXTERNAL" pseudo-headers, like the current "ALL" pseudo-header; these pseudo-headers return all of the headers that were added by either trusted or internal relays and all of the headers that might have been added by either untrusted or external relays (most headers will have been added by untrusted or external relays but there's nothing stopping the trusted or internal relays from adding headers to the bottom of the message); these should be handy for header rules... almost everything that currently uses "ALL" should probably use "ALL-EXTERNAL" or "ALL-UNTRUSTED" - implements support for using the results found in "Received-SPF" headers found in the message; only headers that could have been added by internal relays are used so this is not susceptible to header forgery; "Received-SPF" header results are used by default, there's an option to disable their use; support for both the new and old style of "Received-SPF" headers is included; the new headers support both "mfrom" and "helo" results, the old ones only support "mfrom" AFAIK - added tests to t/spf.t and a new test message to test that the "Received-SPF" headers are only used when they should be (when they're found to be added by an internal relay) and that they're used correctly - slight improvement to the pattern that skips helo's that are IPs and not domain names ------------------------------------------------------------------------ r486633 | jm | 2006-12-13 13:07:13 +0000 (Wed, 13 Dec 2006) | 1 line bug 5242: resurrect previously-removed rules HTML_IMAGE_RATIO_04, HTML_IMAGE_RATIO_06, HTML_IMAGE_RATIO_08 ------------------------------------------------------------------------ r486461 | dos | 2006-12-13 02:25:20 +0000 (Wed, 13 Dec 2006) | 22 lines bug 5236: Support Mail::SPF replacement for Mail::SPF::Query - adds support for Mail::SPF while maintaining fall back support for Mail::SPF::Query; Mail::SPF is the current reference implementation for RFC 4408, it omits a lot of legacy crud present in Mail::SPF::Query - minor optimizations to and removal of crud from the SPF plugin - removes failure comments (for test logs) for all but mfrom or helo FAIL results per RFC 4408 section 6.2 - handles some previously unchecked undef values in debugs - adds tests to t/spf.t for spf based whitelisting (previously untested) - modified t/spf.t to test both of the supported SPF modules specifically - adds an exteremely minimal errorstring() method to M::SA::DnsResolver so that Mail::SPF can use M::SA::DnsResolver instead of its own Net::DNS::Resolver object ------------------------------------------------------------------------ r486134 | jm | 2006-12-12 13:29:38 +0000 (Tue, 12 Dec 2006) | 1 line allow a sleep in the prefork stress test for a new spamd child to start, before attempting to scan; this should fix the intermittent stress-test failures I see here ------------------------------------------------------------------------ r486122 | jm | 2006-12-12 13:03:29 +0000 (Tue, 12 Dec 2006) | 1 line retry listening on spamd sockets 3 times, instead of once, before giving up and issuing an 'address already in use' fatal message. hopefully this may help deal with the spamd_kill_restart.t test failures on the zone with perl 5.6.1 ------------------------------------------------------------------------ r485624 | jm | 2006-12-11 11:42:07 +0000 (Mon, 11 Dec 2006) | 1 line bug 5148: fix Pyzor plugin to support lookups against multiple servers, summing their results. improve Pyzor docs regarding timeouts, and lower the default timeout to 3.5 seconds. thanks to John Hein ------------------------------------------------------------------------ r485479 | felicity | 2006-12-11 03:24:21 +0000 (Mon, 11 Dec 2006) | 1 line bug 5023: add in conf options to update TLD and 2TLD list from conf files ------------------------------------------------------------------------ r485186 | sidney | 2006-12-10 14:17:12 +0000 (Sun, 10 Dec 2006) | 1 line bug 5172: Revert to using gethostbyname (IPv4 only) instead of getaddrinfo with IPv6 support on platforms such as Cygwin that do not have getaddrinfo ------------------------------------------------------------------------ r484837 | dos | 2006-12-08 23:08:47 +0000 (Fri, 08 Dec 2006) | 1 line bug 5234: parse HELO value from Communigate Pro 5.1.3 header ------------------------------------------------------------------------ r483650 | jm | 2006-12-07 20:37:31 +0000 (Thu, 07 Dec 2006) | 1 line bug 5206: detect duplicate rules, and silently merge them internally for greater efficiency. This results in about 100-120KB RAM usage saving in current svn trunk's ruleset, detecting lots of duplicate rules -- so is well worth doing. also, change t/priorities.t so it doesn't accidentally confuse itself with duplicate rules ------------------------------------------------------------------------ r483495 | jm | 2006-12-07 15:33:47 +0000 (Thu, 07 Dec 2006) | 1 line sa-learn used odd --debug-level switch instead of --debug as used everywhere else, and also didn't support debug channels like the other tools do, and as its doco suggests it did; fix, and dup fix in sa-compile too ------------------------------------------------------------------------ r483222 | dos | 2006-12-06 20:49:05 +0000 (Wed, 06 Dec 2006) | 1 line bug 5223: allow for Sendmail's 'may be forged' comment when looking to see if a relay was authenticated ------------------------------------------------------------------------ r483151 | felicity | 2006-12-06 17:00:15 +0000 (Wed, 06 Dec 2006) | 1 line bug 5200: allow 'else' statements ------------------------------------------------------------------------ r483150 | jm | 2006-12-06 17:00:08 +0000 (Wed, 06 Dec 2006) | 1 line bug 4528: fix false positives on Yahoo corporate mail and Yahoo Auto Responses from FORGED_YAHOO_RCVD ------------------------------------------------------------------------ r483138 | felicity | 2006-12-06 16:32:23 +0000 (Wed, 06 Dec 2006) | 1 line bug 5052: update AntiVirus plugin to ignore content-type when looking for exe files ------------------------------------------------------------------------ r482734 | felicity | 2006-12-05 18:43:58 +0000 (Tue, 05 Dec 2006) | 1 line bug 4459: allow 127.0.0.\d+, and look for other illegals ------------------------------------------------------------------------ r482371 | felicity | 2006-12-04 21:37:59 +0000 (Mon, 04 Dec 2006) | 1 line bug 5142: allow decimal numbers, and positive/negative marked numbers in meta rules. when tokenizing, don't require rule names to be over 1 character in length. ------------------------------------------------------------------------ r481873 | felicity | 2006-12-03 20:21:15 +0000 (Sun, 03 Dec 2006) | 1 line bug 5145: better deal with STDIN in spamassassin and sa-learn, since ArchiveIterator doesn't deal with it so well ------------------------------------------------------------------------ r479732 | felicity | 2006-11-27 19:31:22 +0000 (Mon, 27 Nov 2006) | 1 line bug 5210: add back Message::finish() since perl's GC may not completely clean up after our tree, and close non-referenced fd's, etc. ------------------------------------------------------------------------ r473730 | jm | 2006-11-11 15:30:35 +0000 (Sat, 11 Nov 2006) | 1 line bug 5116: 'header RULENAME header:raw' shouldn't strip spaces. ensure it keeps a pristine copy of the header string for matching, and use the test added in the previous checkin (oops ;) ------------------------------------------------------------------------ r473556 | sidney | 2006-11-10 22:50:54 +0000 (Fri, 10 Nov 2006) | 1 line bug 3991: fix testing ssl options in wrong order. --ssl --port=n should disable listening on non-ssl inet port n ------------------------------------------------------------------------ r471778 | jm | 2006-11-06 16:08:05 +0000 (Mon, 06 Nov 2006) | 1 line it should be possible to listen on both SSL and UNIX domain sockets ------------------------------------------------------------------------ r471135 | duncf | 2006-11-04 07:40:56 +0000 (Sat, 04 Nov 2006) | 4 lines Fix logic error when determining whether to create SSL sockets, also a few other SSL related cleanups -- changing $opt{'ssl'} to $listen_ssl now that we can use ssl with --ssl-port and this may not require --ssl. ------------------------------------------------------------------------ r470939 | jm | 2006-11-03 18:40:46 +0000 (Fri, 03 Nov 2006) | 1 line bug 5140: fix parsing of DK tags, tags are case-sensitive, and whitespace may appear before and after tag name and value. let DK_POLICY_TESTING be true if t=y appears in a public key, even when policy record is not fetched. thanks to Mark Martinec ------------------------------------------------------------------------ r470935 | jm | 2006-11-03 18:32:30 +0000 (Fri, 03 Nov 2006) | 1 line bug 5165: 'sa-update --checkonly' added to check for updates without applying them; thanks to ------------------------------------------------------------------------ r470934 | jm | 2006-11-03 18:31:00 +0000 (Fri, 03 Nov 2006) | 1 line bug 3991: spamd can now listen on UNIX domain, TCP, and SSL sockets simultaneously. Command-line semantics extended slightly, although fully backwards compatibly; add the --ssl-port switch to allow TCP and SSL listening at the same time ------------------------------------------------------------------------ r470533 | jm | 2006-11-02 21:39:39 +0000 (Thu, 02 Nov 2006) | 1 line bug 5140: improvements to DomainKeys support, thanks to Mark Martinec: improve debugging; avoid fetching DK policy when the signature is valid (verifies), as required by draft-delany-domainkeys-base-06; avoid sanitizing header (stripping away trailing header fields likely to be inserted by LDA or MUA) when signature header contains a h tag, which explicitly lists header fields which were included in signature calculation. ------------------------------------------------------------------------ r469901 | jm | 2006-11-01 13:35:28 +0000 (Wed, 01 Nov 2006) | 1 line bug 4342: fix hotmail.com FPs on MSGID_FROM_MTA_HEADER, thanks Bob Menschel ------------------------------------------------------------------------ r469477 | parker | 2006-10-31 13:26:22 +0000 (Tue, 31 Oct 2006) | 4 lines Move rule functionality and checking into separate Check plugin. Bug: 4776 ------------------------------------------------------------------------ r467702 | jm | 2006-10-25 16:17:18 +0000 (Wed, 25 Oct 2006) | 1 line bug 5115: make Net::DNS a required module; so many good rules require it that we should try to make it the default ------------------------------------------------------------------------ r465206 | jm | 2006-10-18 10:56:12 +0000 (Wed, 18 Oct 2006) | 1 line bug 5127: allow mimeheader :raw rules to match newlines and folded-header whitespace in MIME header strings ------------------------------------------------------------------------ r465198 | jm | 2006-10-18 10:19:57 +0000 (Wed, 18 Oct 2006) | 1 line add --cf switch to spamassassin, spamd, and sa-learn. This allows the user to provide additional lines of configuration (or rules) directly from the command-line; surprisingly handy ------------------------------------------------------------------------ r464482 | jm | 2006-10-16 13:25:58 +0000 (Mon, 16 Oct 2006) | 1 line fix Sidney's bug; if a meta rule is defined in one rulesrc file, but it depends on another rules defined in other, lexically-later files, and the meta rule is promoted but the dependency rules are not, then the generated code will omit the dependency rule from the active set incorrectly. fix, and add test case ------------------------------------------------------------------------ r464317 | jm | 2006-10-15 22:50:37 +0000 (Sun, 15 Oct 2006) | 1 line bug 4331: change S value to be 0.03, which reduces FPs and FNs at the ends of the Bayes score range, at the cost of increasing 'unsure' cases. since unsure means BAYES_50, no points, we easily make up for those with other rules, so this works better for us ------------------------------------------------------------------------ r452849 | jm | 2006-10-04 11:35:49 +0000 (Wed, 04 Oct 2006) | 1 line bug 4304: in some setups, eg where a plugin implements 'spamd_child_init' with a long-running task, it's possible for child spamd processes to take too long to exit when the parent spamd is killed with SIGINT or SIGTERM. if the caller then starts a new spamd immediately, it will die with 'address already in use', since the old children are still running. Fix this race by unblocking the signals as soon as possible in child spamds after they are forked, and before calling the plugin hook; also add test cases ------------------------------------------------------------------------ r452674 | dos | 2006-10-03 23:21:13 +0000 (Tue, 03 Oct 2006) | 1 line bug 5105: M::SA::Client doesn't always catch failed connection to spamd, fixed ------------------------------------------------------------------------ r450957 | dos | 2006-09-28 18:18:21 +0000 (Thu, 28 Sep 2006) | 1 line bug 5111: fix FORGED_JUNO_RCVD FP on webmail from untd.com's own systems ------------------------------------------------------------------------ r447269 | dos | 2006-09-18 06:01:27 +0000 (Mon, 18 Sep 2006) | 1 line bug 5089: enable adding headers with single digit zero value ------------------------------------------------------------------------ r447014 | felicity | 2006-09-17 04:32:19 +0000 (Sun, 17 Sep 2006) | 1 line do a bunch of work on the Received header parsing code, catch all of my unparseable ham headers, streamline the received header test, add in more headers to test, etc. make more of an API for parse_received_header(), return undef for failed to parse, 0 for parsed and ignored, and a hash reference if parsed. ------------------------------------------------------------------------ r446790 | felicity | 2006-09-15 23:55:22 +0000 (Fri, 15 Sep 2006) | 1 line bug 5101: fix bug in mbx parsing, also update the MBX_SEPARATOR RE to anchor at start and end of lines, which should make things faster. ------------------------------------------------------------------------ r442679 | jm | 2006-09-12 19:31:22 +0000 (Tue, 12 Sep 2006) | 1 line bug 5098: add support for ecelerity Received headers, thanks to Joe Schaefer ------------------------------------------------------------------------ r442013 | felicity | 2006-09-10 21:50:23 +0000 (Sun, 10 Sep 2006) | 1 line bug 4462: allow yes and no to mean 1 and 0 for boolean values ------------------------------------------------------------------------ r441887 | felicity | 2006-09-10 02:08:39 +0000 (Sun, 10 Sep 2006) | 1 line add in a post_config_text option to M::SA::new() which adds the text to the end of the read-in configs, allowing for overrides of things like use_bayes and bayes_path. the previous fix for bug 4206 broke sa-learn options which used to ignore use_bayes. bug 3799: sa-learn --dbpath was accessing the configured bayes_path, then overrode the setting for further use. however, that means we end up touching the configured db before overriding to the one requested. use the new post_config_text to get around that. ------------------------------------------------------------------------ r441496 | jm | 2006-09-08 13:54:00 +0000 (Fri, 08 Sep 2006) | 1 line bug 4928: document max message size limit in spamc; increase from a very arbitrary 9999999 bytes to a much harder-to-hit 250MB; and make it an explicit upper limit in spamc command-line UI. Also, change exit code for this condition to a more appropriate EX_DATAERR if the message is too large ------------------------------------------------------------------------ r440974 | dos | 2006-09-07 04:48:24 +0000 (Thu, 07 Sep 2006) | 1 line bug 5077: fix false SPF_SOFTFAIL's when SPF queries timeout ------------------------------------------------------------------------ r440575 | felicity | 2006-09-06 02:24:57 +0000 (Wed, 06 Sep 2006) | 1 line use temp files for mime parts that we're unlikely to use during processing, and fall-back to memory-only if it's not going to work out. also, use a scalar to hold the decoded information. ------------------------------------------------------------------------ r440453 | felicity | 2006-09-05 19:30:37 +0000 (Tue, 05 Sep 2006) | 1 line bug 5069: add in post_message_parse plugin call, and support for having non-text parts with rendered text. this will allow a plugin for OCR to put the rendered text into the part and have it used automatically in body tests. also, handle certain non-standard content-types, limit internal rendering to just text/plain and text/html. ------------------------------------------------------------------------ r440126 | jm | 2006-09-04 17:22:36 +0000 (Mon, 04 Sep 2006) | 1 line bug 5040: if in no-net scoreset, don't warn about net rules being zero-score dependencies of meta rules ------------------------------------------------------------------------ r440086 | jm | 2006-09-04 14:18:21 +0000 (Mon, 04 Sep 2006) | 1 line memory leak in long-running scan processes using the URIBL rules; their callback closures were not being deleted after each message was scanned. Fixed, and fixed in such a way that plugins don't have to worry about those; they'll just be cleaned up by Mail::SpamAssassin::PerMsgStatus::check() ------------------------------------------------------------------------ r440066 | jm | 2006-09-04 13:18:46 +0000 (Mon, 04 Sep 2006) | 1 line bug 5081: sometimes, SIGHUPing spamd would leave one child process still alive due to a race in the SIGHUP handler and the preforking code. fixed ------------------------------------------------------------------------ r439926 | felicity | 2006-09-04 04:28:16 +0000 (Mon, 04 Sep 2006) | 1 line bug 5076: unescape hash characters in the config ------------------------------------------------------------------------ r439916 | felicity | 2006-09-04 03:45:52 +0000 (Mon, 04 Sep 2006) | 1 line bug 5080: update RCVD_ILLEGAL_IP evaltest to properly deal with 127/8 ------------------------------------------------------------------------ r438116 | jm | 2006-08-29 17:09:31 +0000 (Tue, 29 Aug 2006) | 1 line bug 4603: Mail::SpamAssassin::Spamd::Apache2 -- mod_perl2 module, implementing spamd as a mod_perl module, contributed as a Google Summer of Code project by Radoslaw Zielinski ------------------------------------------------------------------------ r437648 | dos | 2006-08-28 10:56:47 +0000 (Mon, 28 Aug 2006) | 1 line bug 5065: implement whitelist_from_dk ------------------------------------------------------------------------ r436734 | sidney | 2006-08-25 09:24:01 +0000 (Fri, 25 Aug 2006) | 1 line bug 5034: fix endless loop possible from bad input or network error ------------------------------------------------------------------------ r434020 | jm | 2006-08-23 12:04:00 +0000 (Wed, 23 Aug 2006) | 1 line bug 5057: spamc was using getaddrinfo(NULL) to get localhost address info; unfortunately, on linux that returns IPv6 addressing, whereas we want to default to IPv4 unless otherwise specified. use getaddrinfo('127.0.0.1') instead ------------------------------------------------------------------------ r433413 | jm | 2006-08-21 23:01:52 +0000 (Mon, 21 Aug 2006) | 1 line bug 5054: Received-header parsing had differing results based on whether network access was enabled or not; this is inconsistent and a bad idea. reduce everything to the no-net case, for consistency ------------------------------------------------------------------------ r433050 | felicity | 2006-08-20 21:19:03 +0000 (Sun, 20 Aug 2006) | 1 line bug 4848: Pyzor, DCC, and SpamCop plugins weren't looking at the right place to see the 'dont_report_to_...' options ------------------------------------------------------------------------ r431118 | felicity | 2006-08-13 00:20:18 +0000 (Sun, 13 Aug 2006) | 1 line bug 4942: ignore text/calendar parts when looking for rendered/decoded text ------------------------------------------------------------------------ r431056 | felicity | 2006-08-12 17:32:29 +0000 (Sat, 12 Aug 2006) | 1 line don't document looking at local state dir by itself ------------------------------------------------------------------------ r431050 | felicity | 2006-08-12 16:41:43 +0000 (Sat, 12 Aug 2006) | 1 line bug 4952: adding in local_state_dir by itself to the default rules listing means that we'd normally end up looking at something like /var/lib for files, which is bad. change the default local_state_dir to include a 'spamassassin' directory, then modify our previous references to remove the excess 'spamassassin'. ------------------------------------------------------------------------ r430985 | felicity | 2006-08-12 07:03:10 +0000 (Sat, 12 Aug 2006) | 1 line bug 4952: add in DEF_RULES_DIR, LOCAL_RULES_DIR, and LOCAL_STATE_DIR default values if not passed in via M::SA::new(). add documentation for the path macro values passed into M::SA. add local_state_dir by itself as a place to find rules. ------------------------------------------------------------------------ r429908 | sidney | 2006-08-09 01:06:12 +0000 (Wed, 09 Aug 2006) | 1 line Bug 5022: parse Received header generated by command line sendmail ------------------------------------------------------------------------ r428859 | jm | 2006-08-04 19:55:34 +0000 (Fri, 04 Aug 2006) | 1 line bug 3109: simple shortcircuiting, allowing admins to specify that certain rules mean an immediate ham/spam result. thanks to Dallas Engelken ------------------------------------------------------------------------ r427853 | felicity | 2006-08-02 01:53:35 +0000 (Wed, 02 Aug 2006) | 1 line bug 4737: when rewriting a header for markup, strip out leading whitespace before the tag ------------------------------------------------------------------------ r427817 | felicity | 2006-08-02 00:20:50 +0000 (Wed, 02 Aug 2006) | 1 line bug 4492: make bayes_ignore_headers case insensitive ------------------------------------------------------------------------ r427348 | felicity | 2006-07-31 22:27:25 +0000 (Mon, 31 Jul 2006) | 1 line bug 5018: update 2TLD listing in RegistrarBoundaries ------------------------------------------------------------------------ r426739 | felicity | 2006-07-29 01:49:11 +0000 (Sat, 29 Jul 2006) | 1 line bug 5013: handle more IP obfuscation, add in tests, etc. ------------------------------------------------------------------------ r425255 | dos | 2006-07-25 02:00:16 +0000 (Tue, 25 Jul 2006) | 1 line bug 3979: better handling of quoted text when stripping email addresses ------------------------------------------------------------------------ r425244 | felicity | 2006-07-25 01:22:45 +0000 (Tue, 25 Jul 2006) | 1 line bug 5006: more updates for uri redirector obfuscation ------------------------------------------------------------------------ r425095 | felicity | 2006-07-24 15:51:32 +0000 (Mon, 24 Jul 2006) | 1 line bug 5006: canonify obfuscated hostnames in redirection ------------------------------------------------------------------------ r424325 | jm | 2006-07-21 14:06:49 +0000 (Fri, 21 Jul 2006) | 1 line bug 4941: sa-update should defer creating upgrade dir until all files are downloaded, verified, etc. ------------------------------------------------------------------------ r423499 | felicity | 2006-07-19 15:39:02 +0000 (Wed, 19 Jul 2006) | 1 line bug 4997: require Archive::Tar 1.23 and IO::Zlib 1.04 for sa-update ------------------------------------------------------------------------ r423238 | jm | 2006-07-18 20:41:11 +0000 (Tue, 18 Jul 2006) | 1 line bug 4976: increase default message size limit from 250KB to 500KB, to keep up with increasing average spam sizes ------------------------------------------------------------------------ r422963 | felicity | 2006-07-18 04:39:18 +0000 (Tue, 18 Jul 2006) | 1 line bug 4991: if 'spamc -y' is called, and the connection to spamd fails, just print out a blank line and not the whole message ------------------------------------------------------------------------ r421868 | jm | 2006-07-14 10:59:43 +0000 (Fri, 14 Jul 2006) | 1 line bug 3109: priority-tracing -- automatically enforce priority ordering for meta rules, so that metas always run after their dependencies. Remove the now-superfluous META_TEST_MIN_PRIORITY hack. also, change the values of the MISSING_REQUIRED_VALUE and INVALID_VALUE constants in Conf parsing to avoid collisions with the priority range; this is safe as long as no plugins use their raw values, instead of the variable itself, which would be very stupid. ------------------------------------------------------------------------ r421862 | jm | 2006-07-14 10:34:29 +0000 (Fri, 14 Jul 2006) | 1 line bug 4777: evalstr version of run_eval_tests, thanks Michael ------------------------------------------------------------------------ r421289 | jm | 2006-07-12 15:16:17 +0000 (Wed, 12 Jul 2006) | 1 line bug 4966: BSMTP support was broken and would bounce mail with a 421 error; fix by Brad 'anomie' Jorsch ------------------------------------------------------------------------ r420695 | dos | 2006-07-11 03:44:32 +0000 (Tue, 11 Jul 2006) | 1 line bug 4978: fudge out of range times in parse_rfc822_date() to get usable date ------------------------------------------------------------------------ r420676 | dos | 2006-07-11 02:26:51 +0000 (Tue, 11 Jul 2006) | 1 line bug 4980: support Postfix 2.3 auth headers ------------------------------------------------------------------------ r420627 | felicity | 2006-07-10 20:27:48 +0000 (Mon, 10 Jul 2006) | 1 line found another non-text type that occurs in multipart/alternate mails, excluding from MULTIPART_ALT_NON_TEXT rule ------------------------------------------------------------------------ r420382 | felicity | 2006-07-10 00:35:04 +0000 (Mon, 10 Jul 2006) | 1 line bug 4935: fix issue where auto-whitelist db lock file gets the wrong set of permissions ------------------------------------------------------------------------ r418313 | jm | 2006-06-30 15:45:47 +0000 (Fri, 30 Jun 2006) | 1 line bug 4956 comment 8: load IO::Socket::INET ------------------------------------------------------------------------ r418167 | felicity | 2006-06-30 03:01:29 +0000 (Fri, 30 Jun 2006) | 1 line bug 4921: localize spamd rc script messages for redhat distros ------------------------------------------------------------------------ r418149 | jm | 2006-06-29 22:22:38 +0000 (Thu, 29 Jun 2006) | 1 line cleanup spamc error handling and reduce ifdefs; fix a windows build failure due to misplaced format args in libspamc_log; log hostname in failure-to-connect messages again; log failures to connect for all failing interfaces for multihomed hosts, so that t/spamc_l.t will pass when there is no IPv6 interface but IPv6 is in the kernel ------------------------------------------------------------------------ r418135 | jm | 2006-06-29 20:56:58 +0000 (Thu, 29 Jun 2006) | 10 lines some changes in response to comment 14 in bug 4477. - use NI_NUMERICSERV to list port numbers instead of names; - use NI_MAXHOST and NI_MAXSERV where available; - use getaddrinfo()'s built-in support for resolving the loopback address, without requiring a resolver. however the max-hosts issue and the _randomize_hosts() issue are still outstanding; not biggies though I think. ------------------------------------------------------------------------ r418122 | jm | 2006-06-29 19:47:32 +0000 (Thu, 29 Jun 2006) | 1 line possible fix for bug 4477 coredump ------------------------------------------------------------------------ r418104 | jm | 2006-06-29 18:10:22 +0000 (Thu, 29 Jun 2006) | 1 line bug 4956: IPv6 support in spamd, thanks to Radoslaw Zielinski ------------------------------------------------------------------------ r418101 | jm | 2006-06-29 17:58:02 +0000 (Thu, 29 Jun 2006) | 1 line bug 4477: port spamc to support both IPv4 and IPv6; thanks to John Madden ------------------------------------------------------------------------ r417337 | felicity | 2006-06-27 04:34:25 +0000 (Tue, 27 Jun 2006) | 1 line bug 4488: deal with a circular reference between Bayes and BayesStore -- when M::SA::finish() is called, make sure to delete the Bayes->BayesStore reference, which will then let everything else get cleaned up as expected. ------------------------------------------------------------------------ r417315 | felicity | 2006-06-27 00:11:12 +0000 (Tue, 27 Jun 2006) | 1 line bug 4968: allow % in paths when untainting ------------------------------------------------------------------------ r417119 | felicity | 2006-06-26 03:28:43 +0000 (Mon, 26 Jun 2006) | 1 line bug 4969: make the recursive option to find_parts() useful by allowing recursion into direct children but not go further, fixing the old version which only looked at the single part ------------------------------------------------------------------------ r416977 | dos | 2006-06-25 02:11:16 +0000 (Sun, 25 Jun 2006) | 1 line bug 4943: parse Norton AntiVirus Gateway received headers ------------------------------------------------------------------------ r415638 | jm | 2006-06-20 13:39:20 +0000 (Tue, 20 Jun 2006) | 1 line reverse de-tainting of uid/gid from getpwnam() calls, needless. see dev traffic under 'untainting getpwnam output' thread ------------------------------------------------------------------------ r415397 | jm | 2006-06-19 19:49:47 +0000 (Mon, 19 Jun 2006) | 1 line bug 4926: given a certain set of parameters to spamd and a specially formatted input message, users could cause spamd to execute arbitrary commands as spamd user ------------------------------------------------------------------------ r410099 | parker | 2006-05-29 14:50:55 +0000 (Mon, 29 May 2006) | 1 line Bug 4825: Fix silly error trying to match just PONG on the return ------------------------------------------------------------------------ r410038 | dos | 2006-05-29 07:06:09 +0000 (Mon, 29 May 2006) | 1 line bug 4924: initialize tag data for _DCCB_, _DCCR_, _PYZOR_, _RBL_, _BAYESTC*_ ------------------------------------------------------------------------ r410004 | parker | 2006-05-28 23:39:03 +0000 (Sun, 28 May 2006) | 1 line Bug 4902: Make multiple TextCat settings changable via configuration ------------------------------------------------------------------------ r406998 | jm | 2006-05-16 17:29:52 +0000 (Tue, 16 May 2006) | 1 line bug 4901: fix 'dns: sendto() failed' error, fix thanks to Mark Martinec ------------------------------------------------------------------------ r399368 | felicity | 2006-05-03 17:58:52 +0000 (Wed, 03 May 2006) | 1 line bug 4888: add in a _HEADER(...)_ tag to include information from headers ------------------------------------------------------------------------ r399363 | felicity | 2006-05-03 17:49:05 +0000 (Wed, 03 May 2006) | 1 line add in a SUBTESTS tag that was requested on the users list ------------------------------------------------------------------------ r398792 | felicity | 2006-05-02 03:38:37 +0000 (Tue, 02 May 2006) | 1 line bug 4870: add in rc script support for DragonFlyBSD ------------------------------------------------------------------------ r398277 | felicity | 2006-04-30 03:39:28 +0000 (Sun, 30 Apr 2006) | 1 line bug 4884: if no message content is passed in, fake a message with only a single newline to avoid undef warning issues ------------------------------------------------------------------------ r396242 | sidney | 2006-04-23 08:50:39 +0000 (Sun, 23 Apr 2006) | 1 line bug 4874: userconf rules were not being made active, for example GTUBE ------------------------------------------------------------------------ r396213 | sidney | 2006-04-23 04:18:54 +0000 (Sun, 23 Apr 2006) | 1 line bug 4861: dcc code cleanup, add use of dccifd for reporting, set default timeout to 8 seconds, add test case ------------------------------------------------------------------------ r396121 | jm | 2006-04-22 13:11:04 +0000 (Sat, 22 Apr 2006) | 1 line bug 4326: add spamc --long-option support, thanks to John Madden ------------------------------------------------------------------------ r395391 | jm | 2006-04-19 20:47:19 +0000 (Wed, 19 Apr 2006) | 1 line bug 4630: fix HP-UX 11.11 libspamc build, patch from John Madden ------------------------------------------------------------------------ r394348 | jm | 2006-04-15 18:40:50 +0000 (Sat, 15 Apr 2006) | 1 line bug 4860: consolidate DNS-lookup async-rule infrastructure, increasing code-sharing between URIDNSBL.pm and Dns.pm and allowing 19% speedup ------------------------------------------------------------------------ r394297 | jm | 2006-04-15 14:58:13 +0000 (Sat, 15 Apr 2006) | 1 line bug 4867: fetchmail 6.3.2.1-1.2 apparently changed received header format, triggering (harmless) UNPARSEABLE_RELAY hits ------------------------------------------------------------------------ r393884 | felicity | 2006-04-13 18:53:37 +0000 (Thu, 13 Apr 2006) | 1 line make parsing of style="..." generic to any tag (still limited by ok_attributes), move strange else { if ... } to just be elsif {}, strip out unnecessary variable, etc. ------------------------------------------------------------------------ r393832 | felicity | 2006-04-13 15:26:19 +0000 (Thu, 13 Apr 2006) | 1 line style="visibility: hidden" ought to be caught as making text invisible... ------------------------------------------------------------------------ r393358 | dos | 2006-04-12 01:05:42 +0000 (Wed, 12 Apr 2006) | 1 line bug 3838: work around Perl bug causing to become tainted -- thanks to Mark Martinec for pointing out the bug with Perl itself ------------------------------------------------------------------------ r392709 | sidney | 2006-04-09 12:53:21 +0000 (Sun, 09 Apr 2006) | 1 line bug 4793: don't substitute anything for undefined template tags ------------------------------------------------------------------------ r392533 | sidney | 2006-04-08 14:35:44 +0000 (Sat, 08 Apr 2006) | 1 line bug 4828: pass arguments to dccifd based on what we calculate as the proper Received header instead of allowing dccifd to always use the first Received header ------------------------------------------------------------------------ r392480 | sidney | 2006-04-08 04:58:12 +0000 (Sat, 08 Apr 2006) | 1 line bug 4859: plug memory leak. Even though it was released when child ended it limited the practical max number of children ------------------------------------------------------------------------ r392082 | dos | 2006-04-06 21:22:29 +0000 (Thu, 06 Apr 2006) | 1 line bug 3871: add option to spamd to specify group to run as ------------------------------------------------------------------------ r391838 | jm | 2006-04-05 22:47:23 +0000 (Wed, 05 Apr 2006) | 1 line bug 3815: fix RELAYCOUNTRY tag so that it really works this time ------------------------------------------------------------------------ r391755 | dos | 2006-04-05 19:46:22 +0000 (Wed, 05 Apr 2006) | 1 line bug 4852: race condition in spamd pre-fork causes harmless warnings in logs, fixed ------------------------------------------------------------------------ r391287 | jm | 2006-04-04 12:23:25 +0000 (Tue, 04 Apr 2006) | 1 line bug 4767: all net checks will stop forever if is_dns_available() encounters a single failure; add 'dns_test_interval' instead to auto-retest every 10 minutes, patch from Dallas Engelken ------------------------------------------------------------------------ r391158 | dos | 2006-04-03 22:20:18 +0000 (Mon, 03 Apr 2006) | 1 line bug 4699: avoid cases where timeouts can occur without a die() call ------------------------------------------------------------------------ r391067 | jm | 2006-04-03 15:18:03 +0000 (Mon, 03 Apr 2006) | 1 line bug 4593: revert changes made for bug 3452, where some spamc vars were made unsigned, resulting in infinite loop caused by an error condition in spamc ------------------------------------------------------------------------ r390942 | felicity | 2006-04-03 02:13:24 +0000 (Mon, 03 Apr 2006) | 1 line bug 4850: reenable the razor plugin because of the usage license change as of 3/13/2006 ------------------------------------------------------------------------ r388666 | dos | 2006-03-24 22:30:17 +0000 (Fri, 24 Mar 2006) | 1 line bug 4802: whitelisting: use optional indentity instead of signer since identities can be stricter but encompass the signer domain ------------------------------------------------------------------------ r388402 | dos | 2006-03-24 06:18:35 +0000 (Fri, 24 Mar 2006) | 1 line bug 4802: implement whitelist_from_dkim ------------------------------------------------------------------------ r387224 | jm | 2006-03-20 15:47:29 +0000 (Mon, 20 Mar 2006) | 1 line bug 4824: 'tryplugin' keyword, which allows sandbox plugins to be supported cleanly in mkrules ------------------------------------------------------------------------ r385960 | felicity | 2006-03-15 03:27:34 +0000 (Wed, 15 Mar 2006) | 1 line bug 3754: cause AI::run() to return a value indicating if there was a problem, and let sa-learn see this and set the exit code to non-zero if there was a problem (such as a file wasn't able to be opened). also, we used die() in places where we should have been able to recover, so throw a warning instead and recover. ------------------------------------------------------------------------ r385943 | felicity | 2006-03-15 00:43:36 +0000 (Wed, 15 Mar 2006) | 1 line bug 4813: avoid misparsing sendmail received header as qmail ------------------------------------------------------------------------ r384629 | felicity | 2006-03-09 22:04:52 +0000 (Thu, 09 Mar 2006) | 1 line bug 4822: change rawbody to run over the entire undecoded part content, not split into lines or paragraphs ------------------------------------------------------------------------ r383976 | sidney | 2006-03-07 19:40:14 +0000 (Tue, 07 Mar 2006) | 1 line use size_t instead of socket_t for compatibility with Win32 some old FreeBSD and maybe other platforms ------------------------------------------------------------------------ r383701 | sidney | 2006-03-06 23:19:08 +0000 (Mon, 06 Mar 2006) | 1 line Bug 3710: fix spamc to timeout as -t option specifies when tcp and unix connects fail. Also fix some debugging code normally not compiled ------------------------------------------------------------------------ r383466 | dos | 2006-03-06 05:08:31 +0000 (Mon, 06 Mar 2006) | 6 lines - provide for $pms->{relays_internal} and $pms->{relays_external} and their related tags and metadata - bug 4760 makes this possible - this allows for simpler determination of the proper external relay to test when doing SPF checks, etc. ------------------------------------------------------------------------ r381804 | dos | 2006-02-28 22:44:48 +0000 (Tue, 28 Feb 2006) | 1 line bug 4760: don't check to see if excluded internal_networks are included in trusted_networks && don't match included internal_networks in excluded trusted_networks ------------------------------------------------------------------------ r381731 | felicity | 2006-02-28 18:36:40 +0000 (Tue, 28 Feb 2006) | 1 line in set_bool_value, if the value is actually a string, perl throws a warning. use a RE instead. ------------------------------------------------------------------------ r381552 | felicity | 2006-02-28 03:51:33 +0000 (Tue, 28 Feb 2006) | 1 line bug 4090: x86_64 platforms (linux specifically) seem to have an issue with compiling libspamc.so, so disable it for now since it's not necessary for our stuff ------------------------------------------------------------------------ r380783 | jgmyers | 2006-02-24 18:53:34 +0000 (Fri, 24 Feb 2006) | 1 line Bugs 4606, 4609: Adjust MIME parsing limits ------------------------------------------------------------------------ r380241 | jm | 2006-02-23 22:19:29 +0000 (Thu, 23 Feb 2006) | 1 line bug 4760: hosts in internal_networks should always be in trusted_networks; one is a subset of the other. enforce this and make the error condition an explicit error ------------------------------------------------------------------------ r379942 | felicity | 2006-02-22 22:19:59 +0000 (Wed, 22 Feb 2006) | 1 line bug 4422: validate pyzor and dcc paths ------------------------------------------------------------------------ r379681 | felicity | 2006-02-22 04:54:55 +0000 (Wed, 22 Feb 2006) | 1 line bug 4254: don't allow '**' or '* *' in meta expressions ------------------------------------------------------------------------ r379629 | jm | 2006-02-21 23:56:07 +0000 (Tue, 21 Feb 2006) | 1 line bug 4791: avoid 'Malformed UTF-8' warnings caused by bad UTF data in the message and a tr/// statement used to measure body string length ------------------------------------------------------------------------ r379281 | jm | 2006-02-20 23:34:15 +0000 (Mon, 20 Feb 2006) | 1 line bug 4790: sa-update should not use HOME/.gnupg for its keyring. instead, use /etc/mail/spamassassin/sa-update-keys ------------------------------------------------------------------------ r378959 | felicity | 2006-02-19 21:40:08 +0000 (Sun, 19 Feb 2006) | 1 line if a config line was in a conditional, and another related line came along, the conditional could be erased (ie: score FOO 0, lang de describe FOO blah) ------------------------------------------------------------------------ r378437 | felicity | 2006-02-17 05:23:07 +0000 (Fri, 17 Feb 2006) | 1 line bug 4800: fix the RE parsing X-Spam-Status headers during mass-check net reuse ------------------------------------------------------------------------ r377834 | dos | 2006-02-14 21:19:04 +0000 (Tue, 14 Feb 2006) | 1 line bug 4795: validate score config option ------------------------------------------------------------------------ r376189 | felicity | 2006-02-09 05:07:18 +0000 (Thu, 09 Feb 2006) | 1 line bug 4695: fix RE to just replace '' instead of looking at whole tags. ------------------------------------------------------------------------ r374739 | dos | 2006-02-03 20:50:43 +0000 (Fri, 03 Feb 2006) | 1 line bug 4782: add domainkeys_timeout config option ------------------------------------------------------------------------ r374732 | dos | 2006-02-03 19:42:53 +0000 (Fri, 03 Feb 2006) | 1 line bug 4781: add spf_timeout config option ------------------------------------------------------------------------ r374643 | dos | 2006-02-03 11:54:36 +0000 (Fri, 03 Feb 2006) | 1 line bug 4780: fix IP_ADDRESS & LOCALHOST regexes to correctly parse IPv6 addresses ------------------------------------------------------------------------ r373061 | jm | 2006-01-28 01:31:31 +0000 (Sat, 28 Jan 2006) | 1 line bug 4696: eradicate spurious 'tcp timeout' and 'select returned error on server filehandle' warnings from SpamdForkScaling ------------------------------------------------------------------------ r372567 | jgmyers | 2006-01-26 16:51:03 +0000 (Thu, 26 Jan 2006) | 1 line Bug 4636: Make charset normalization enabled by the normalize_charset config option ------------------------------------------------------------------------ r372080 | jm | 2006-01-25 02:48:48 +0000 (Wed, 25 Jan 2006) | 1 line bug 4695: use of HTML::Parser->empty_element_tags() allowed a number of HTML obfuscations past, demonstrated by uri.t test failures. ignore it and instead 'hack' fix with a simple s/// substitution ------------------------------------------------------------------------ r370724 | dos | 2006-01-20 04:00:40 +0000 (Fri, 20 Jan 2006) | 1 line prevent trusted/internal network settings that will have no effect; this should prevent people from getting confused when setting inclusion/exclusion networks/ips ------------------------------------------------------------------------ r370351 | jm | 2006-01-19 03:19:05 +0000 (Thu, 19 Jan 2006) | 1 line don't look for 'languages', 'triplets.txt' and 'user_prefs.template' in the /var/lib/spamassassin/3.002000 directory; and don't ship them there either in the update tarball. (see 'updates being published' thread on dev list) ------------------------------------------------------------------------ r368197 | jm | 2006-01-12 01:04:50 +0000 (Thu, 12 Jan 2006) | 1 line use exit code 1 if no updates are available, so that callers can tell the difference between 'updates installed, spamd restart needed' and 'no updates installed' ------------------------------------------------------------------------ r367853 | jm | 2006-01-11 00:59:08 +0000 (Wed, 11 Jan 2006) | 1 line bug 4759: avoid allowing spammers to hide a relay line using a faked fetchmail handover ------------------------------------------------------------------------ r366594 | jgmyers | 2006-01-06 23:09:10 +0000 (Fri, 06 Jan 2006) | 1 line Bug 4636: Add support for charset normalization ------------------------------------------------------------------------ r357663 | jm | 2005-12-19 07:47:59 +0000 (Mon, 19 Dec 2005) | 1 line bug 4623: use Mail::DomainKeys::Message::signature() API if it exists, otherwise fall back to the old API ------------------------------------------------------------------------ r357659 | jm | 2005-12-19 07:23:50 +0000 (Mon, 19 Dec 2005) | 1 line bug 4611: use lookup_ptr() when looking up an IP address, as Net::DNS::Resolver does ------------------------------------------------------------------------ r356946 | jm | 2005-12-14 23:54:19 +0000 (Wed, 14 Dec 2005) | 1 line bug 4735: lack of network interfaces causes 'spamassassin -L' to crash ------------------------------------------------------------------------ r356646 | jgmyers | 2005-12-13 23:29:16 +0000 (Tue, 13 Dec 2005) | 1 line Bug 3787: Bump HTML::Parser minimum version to prevent mailformed UTF-8 errors ------------------------------------------------------------------------ r356300 | dos | 2005-12-12 17:48:10 +0000 (Mon, 12 Dec 2005) | 2 lines bug 4728: add foo-lastexternal code for DUL checks also: fix dbg message in check_rbl_backend() ------------------------------------------------------------------------ r355362 | jm | 2005-12-09 04:35:56 +0000 (Fri, 09 Dec 2005) | 1 line bug 3815: allow 'add_header all Relay _RELAYCOUNTRY_' to work with RelayCountry plugin ------------------------------------------------------------------------ r355211 | jm | 2005-12-08 21:19:48 +0000 (Thu, 08 Dec 2005) | 1 line bug 4617: fix for spurious 'an't call method finish on an undefined value' errors in AWL plugin ------------------------------------------------------------------------ r355190 | felicity | 2005-12-08 20:17:19 +0000 (Thu, 08 Dec 2005) | 1 line bug 4719: try to be nice when splitting long body lines, look for a space ------------------------------------------------------------------------ r354530 | felicity | 2005-12-06 20:05:40 +0000 (Tue, 06 Dec 2005) | 1 line bug 4722: if we've already loaded a plugin via path specification, don't bother trying to load it again (causes warnings when functions get redefined) ------------------------------------------------------------------------ r351875 | felicity | 2005-12-03 02:38:23 +0000 (Sat, 03 Dec 2005) | 1 line bug 4485: validate IP before calling my_inet_aton() ------------------------------------------------------------------------ r350102 | jm | 2005-12-01 02:12:36 +0000 (Thu, 01 Dec 2005) | 1 line bug 4700: certain privileged configuration settings can inject code, due to a bad fix for bug 3846. Back that out ------------------------------------------------------------------------ r350101 | jm | 2005-12-01 02:06:10 +0000 (Thu, 01 Dec 2005) | 1 line bug 4590: add some workaround code for the 'syswrite(6) failed, retrying...' prefork bug; instead of retrying indefinitely, it'll retry 3 times then kill the child and allow it to respawn ------------------------------------------------------------------------ r349500 | jm | 2005-11-28 20:50:28 +0000 (Mon, 28 Nov 2005) | 1 line bug 4631: __DIE__ hooks should not be called for timeout/exception-handling events; use the perlfunc-recommended workaround ------------------------------------------------------------------------ r349405 | sidney | 2005-11-28 09:30:23 +0000 (Mon, 28 Nov 2005) | 1 line bug 4707 update. faster fix than first patch, fixes another buffer overflow from malformed input. Also comment out unused const that caused compiler warning in MacOS X ------------------------------------------------------------------------ r349193 | sidney | 2005-11-27 08:41:55 +0000 (Sun, 27 Nov 2005) | 1 line bug 4708: properly parse bad spamd token ------------------------------------------------------------------------ r349191 | sidney | 2005-11-27 07:46:39 +0000 (Sun, 27 Nov 2005) | 1 line bug 4707: Fix seg fault when spamc given short invalid input with -B option, add test case for it ------------------------------------------------------------------------ r348370 | sidney | 2005-11-23 04:51:34 +0000 (Wed, 23 Nov 2005) | 1 line bug 4704: Properly parse Received headers such as those generated by Nemesis ------------------------------------------------------------------------ r348079 | felicity | 2005-11-22 04:46:42 +0000 (Tue, 22 Nov 2005) | 1 line bug 4698: allow 'meta RULE 0' to work ------------------------------------------------------------------------ r345454 | jm | 2005-11-18 05:47:33 +0000 (Fri, 18 Nov 2005) | 1 line bug 4655: create .pid file for spamassassin service to avoid killing the wrong processes and leaving spamd running ------------------------------------------------------------------------ r345091 | jgmyers | 2005-11-16 18:59:20 +0000 (Wed, 16 Nov 2005) | 1 line Bug 4349: header rules now support tflags multiple ------------------------------------------------------------------------ r345076 | jgmyers | 2005-11-16 18:07:21 +0000 (Wed, 16 Nov 2005) | 1 line Bug 4349: meta and eval rules can now also have counts ------------------------------------------------------------------------ r344480 | jgmyers | 2005-11-15 23:46:50 +0000 (Tue, 15 Nov 2005) | 1 line Bug 4349: implement tflags multiple to allow eval rules to count subtest hits ------------------------------------------------------------------------ r344247 | dos | 2005-11-14 22:01:12 +0000 (Mon, 14 Nov 2005) | 1 line bug 4686: parse obtuse smtpd received headers ------------------------------------------------------------------------ r332728 | dos | 2005-11-12 04:02:32 +0000 (Sat, 12 Nov 2005) | 1 line bug 4627: parse University of Michigan simpta SMTP headers ------------------------------------------------------------------------ r332721 | jm | 2005-11-12 03:03:31 +0000 (Sat, 12 Nov 2005) | 1 line add support for --after switch, to reduce the massive volumes of mass-checkable mail ------------------------------------------------------------------------ r332709 | dos | 2005-11-12 02:10:45 +0000 (Sat, 12 Nov 2005) | 1 line bug 4680: detect Courier SMTP auth tokens ------------------------------------------------------------------------ r332694 | jm | 2005-11-12 00:59:52 +0000 (Sat, 12 Nov 2005) | 1 line major revamp; I was trying to avoid using IPC::DirQueue, but it just makes a lot more sense ------------------------------------------------------------------------ r332684 | jm | 2005-11-12 00:28:56 +0000 (Sat, 12 Nov 2005) | 1 line add SIGINT handler, pidfile, and --wake switches ------------------------------------------------------------------------ r332683 | dos | 2005-11-12 00:28:49 +0000 (Sat, 12 Nov 2005) | 1 line bug 4631: catch eval'd die's leaked by broken perl __DIE__ hooks ------------------------------------------------------------------------ r331697 | dos | 2005-11-08 05:51:28 +0000 (Tue, 08 Nov 2005) | 1 line bug 4672: Add syntax to exclude hosts/networks from trusted/internal networks ------------------------------------------------------------------------ r331692 | dos | 2005-11-08 05:38:37 +0000 (Tue, 08 Nov 2005) | 1 line bug 4661: SPF checks should be done against trusted, but external relays ------------------------------------------------------------------------ r331591 | jgmyers | 2005-11-07 19:59:53 +0000 (Mon, 07 Nov 2005) | 1 line Bug 4517: instrument eval rules as well ------------------------------------------------------------------------ r331590 | jgmyers | 2005-11-07 19:52:15 +0000 (Mon, 07 Nov 2005) | 1 line Bug 4515: content preview omits first paragraph when no Subject: header ------------------------------------------------------------------------ r328495 | sidney | 2005-10-25 22:36:02 +0000 (Tue, 25 Oct 2005) | 1 line bug 4619 update to work with older versions of Net::DNS less than 0.50 ------------------------------------------------------------------------ r327897 | felicity | 2005-10-24 01:04:30 +0000 (Mon, 24 Oct 2005) | 1 line bug 4363: if a message uses CRLF for line endings, we should use it as well, otherwise stay with LF as usual ------------------------------------------------------------------------ r327648 | sidney | 2005-10-22 09:51:31 +0000 (Sat, 22 Oct 2005) | 1 line Bug 4639. Add -K option to spamc to ping spamd ------------------------------------------------------------------------ r327626 | jm | 2005-10-22 03:10:27 +0000 (Sat, 22 Oct 2005) | 1 line cope with usernames containing dashes; better coping with multiple hit-freqs processes going in parallel ------------------------------------------------------------------------ r326275 | jm | 2005-10-18 22:15:22 +0000 (Tue, 18 Oct 2005) | 1 line moving all these files back again, from rulescode to rules. rulescode is no more ------------------------------------------------------------------------ r326274 | jm | 2005-10-18 22:12:52 +0000 (Tue, 18 Oct 2005) | 1 line ignore automc buildbot scripts for packaging; also, ignore rules project source files, which are SVN-only ------------------------------------------------------------------------ r326272 | jm | 2005-10-18 22:11:28 +0000 (Tue, 18 Oct 2005) | 21 lines change of plan -- let the wookie win. ;) Avoid MANIFEST/"dist"/install issues, by limiting the rule filenames to just a pre-defined set. That way, we're back to MANIFEST containing the list of files in the "rules" directory. That dir includes both code-tied (source) and rules project (built) rules files. The "rulescode" dir is gone, again. The "rulesrc" dir is SVN-only, and not in tarballs -- the package produced by make dist/make tardist etc. will contain the built rules files in "rules". "make install" then installs those files. This means that there may still be issues running "make install" from an SVN checkout -- but it will always work from a tarball, since that will contain the "built" files. Non-"publish"ed sandbox rules are written to rules/70_sandbox.cf, and this file is *not* installed. ------------------------------------------------------------------------ r325836 | sidney | 2005-10-17 03:56:55 +0000 (Mon, 17 Oct 2005) | 1 line Bug 4619 update fix to include test for INET6 socket with ipv4 addresses, and to remove need to refer to AF_INET6 macro by seting PeerAddr instead of Domain in call to socket new ------------------------------------------------------------------------ r322462 | sidney | 2005-10-16 09:37:20 +0000 (Sun, 16 Oct 2005) | 1 line Bug 4619 - Disable use of IPv6 for DNS if unable to create INET6 socket and add --ipv4 command line option to use in case there is still a problem ------------------------------------------------------------------------ r315047 | jgmyers | 2005-10-12 20:59:47 +0000 (Wed, 12 Oct 2005) | 1 line bug 4596: Normalize Unicode HTML entities. Increases minimum version of HTML::Parser to 3.43 ------------------------------------------------------------------------ r294971 | jm | 2005-10-05 02:05:48 +0000 (Wed, 05 Oct 2005) | 1 line bug 4517: rule instrumentation plugin hooks, from John Gardiner Myers ------------------------------------------------------------------------ r293487 | jm | 2005-10-04 02:02:35 +0000 (Tue, 04 Oct 2005) | 1 line bug 4597: avoid stupid 'can't call method 'string' on an undefined value' errors from DnsResolver ------------------------------------------------------------------------ r290904 | dos | 2005-09-22 07:37:12 +0000 (Thu, 22 Sep 2005) | 1 line bug 4425: whitelist_from_rcvd shouldn't look at trusted headers if untrusted ones exist ------------------------------------------------------------------------ r289976 | felicity | 2005-09-18 21:17:52 +0000 (Sun, 18 Sep 2005) | 1 line bug 4589: allow M::SA::Message to use IO::File objects to read in message (same as GLOB). Also add in check to ignore unknown reference types. ------------------------------------------------------------------------ r280738 | jm | 2005-09-14 01:53:11 +0000 (Wed, 14 Sep 2005) | 1 line bug 4522: ISO-2022 charset shifting escape sequence threw off bare-URI parsing. fixed ------------------------------------------------------------------------ r280056 | jm | 2005-09-11 01:28:07 +0000 (Sun, 11 Sep 2005) | 1 line bug 4565: deal with new Yahoo hostname convention which caused FORGED_YAHOO_RCVD fps ------------------------------------------------------------------------ r279666 | sidney | 2005-09-08 22:51:18 +0000 (Thu, 08 Sep 2005) | 1 line Bug 4570 fix regexp to handle longer folded headers without segfaulting ------------------------------------------------------------------------ r279516 | jm | 2005-09-08 05:11:06 +0000 (Thu, 08 Sep 2005) | 1 line bug 4572: catch redirections using EmailFactory (emf0.com) ------------------------------------------------------------------------ r264935 | dos | 2005-08-31 04:01:31 +0000 (Wed, 31 Aug 2005) | 1 line bug 4558: add redirector pattern for oscommerce software open redirector ------------------------------------------------------------------------ r263807 | duncf | 2005-08-28 03:31:05 +0000 (Sun, 28 Aug 2005) | 2 lines Bug 4494: Fix sa-learn's --local option ------------------------------------------------------------------------ r261908 | duncf | 2005-08-28 01:04:10 +0000 (Sun, 28 Aug 2005) | 2 lines Bug 4535: tweak MIME boundary regexp ------------------------------------------------------------------------ r234434 | duncf | 2005-08-22 04:44:01 +0000 (Mon, 22 Aug 2005) | 3 lines Bug 4546: disable TELL commands by default, add command line option --allow-tell or -l to enable. ------------------------------------------------------------------------ r233991 | jm | 2005-08-20 03:03:34 +0000 (Sat, 20 Aug 2005) | 1 line bug 4518: fix situation where helper apps can be left running after SA times out on them, thanks to Mark Martinec ------------------------------------------------------------------------ r233987 | jm | 2005-08-20 02:56:36 +0000 (Sat, 20 Aug 2005) | 1 line bug 4533: syslog entries were being identified as 'spamassassin' instead of 'spamd'. fixed, thanks to Dawn Keenan ------------------------------------------------------------------------ r233986 | jm | 2005-08-20 02:43:15 +0000 (Sat, 20 Aug 2005) | 1 line bug 4412: try to deal with platforms that don't yet have AF_INET6 ------------------------------------------------------------------------ r233398 | jm | 2005-08-18 23:44:56 +0000 (Thu, 18 Aug 2005) | 1 line bug 4412: DNS resolver fails with an IPv6 resolver address in /etc/resolv.conf; add IPv6 support code ------------------------------------------------------------------------ r233111 | jm | 2005-08-17 03:09:43 +0000 (Wed, 17 Aug 2005) | 1 line bug 4536: avoid breakage when warns happen before Mail::SA::Logger is loaded ------------------------------------------------------------------------