* Add counter for oversized chunks: this parameter can improve performance, for example
if we have too many oversized chunks it is good decision to increase
chunk size.
Vsevolod Stakhov [Mon, 30 Mar 2009 13:57:59 +0000 (17:57 +0400)]
* Add message handling functions to lua API
* Add ability to add lua code in config with .lua and .endlua
* Add consolidation functions support that are written in perl or lua
Vsevolod Stakhov [Sun, 29 Mar 2009 10:02:01 +0000 (14:02 +0400)]
* Make regexp cache functions public for using them in other places
* Add function regexp_match_number that checks number of matched arguments:
e.g.: regexp_match_number(2, ${__RE1}, ${__RE2}, header_exists(Subject))
Vsevolod Stakhov [Mon, 23 Mar 2009 11:10:07 +0000 (14:10 +0300)]
* Fix error in expression parser that causes bad errors with expressions that have regexp at the end
* Improve test for fuzzy hashes
* Add new object - TextPart to perl XS library that allows access to stripped parts and fuzzy hashes
* Add documentation for expressions parser and fot Mail::Rspamd::TextPart
* Allways calculate fuzzy hash for text parts
* Store text parts separately from other parts
* Add compare_parts_distance for expressions that calculates difference in 2 parts messages
* Do not try to substitute variables in empty strings
Vsevolod Stakhov [Thu, 19 Mar 2009 14:44:57 +0000 (17:44 +0300)]
* Add functions support to rspamd regexps
* Parse expressions with state machine which allows different kinds of arguments in expressions
* Fix test to accord current data
* Add support of fucntions to regexp module
* Move all regexp logic to separate file, describe its API
* Fix descriptors leakage in surbl module
Vsevolod Stakhov [Thu, 12 Mar 2009 13:39:52 +0000 (16:39 +0300)]
* Add sections support to rspamd statfiles API
* Change logic of statfile pool:
- statfiles hash is no longer shared hash as we have per-process uniq things like current section or offset in statfile
- introduce shared hash of statfiles mmapped areas to avoid multiply mmaps of the same file
Vsevolod Stakhov [Wed, 11 Mar 2009 09:08:11 +0000 (12:08 +0300)]
* Fixes to redirector:
- move memcached servers setting to config hash
- eval config file before any actions
- move daemonization code after all subs
- fix searching in domains for redirector's check
Vsevolod Stakhov [Tue, 10 Mar 2009 16:54:20 +0000 (19:54 +0300)]
* Fix redirector connection procedure
* Add more strict login
* Add new header Queue-ID to protocol
* Log message id or queue id
* Add config file for redirector
* Add ability to set regexp and domains list to check with redirector
Vsevolod Stakhov [Tue, 10 Mar 2009 11:20:14 +0000 (14:20 +0300)]
* Fix 2 memory issues:
- NULL string when trying to check url regexp
- double free of message byte array as mime_stream frees memory in array if it thinks that stream is
owner of array's memory
* Fix bug in url parser with freeing memory allocated by memory_pool allocator
* Do not write to log if we don't do any read operation in dispatcher library
* Add ability to fork multiply lmtp workers
* Add ability to drop privilleges of rspamd processes
* Add ability to install redirector with rspamd
* Add FreeBSD start scripts for rspamd and redirector
* Remove command line parsing from util.c as it has specific usage only in main process
* Unify socket creation and accepting by using utility functions
* Check SO_ERROR on socket immideately after connect/bind to avoid resourses allocation for
unsuccessful connections
* Avoid descriptors leakage when we create sockets and have errors on them
* Set on all socket descriptors FD_CLOEXEC flag to avoid problems with executing LDA
* New rspamd protocol (changed replies)
* Add ability to attach string lists to symbols
* Check destructors and do not add identical destructors (argument and function) to pool's destructors
* Remove 2 warnings when building with gmime22
* Attach url names to surbl symbols
* Fix bug with blocking on read (I think it is linux specific thought)