mailman.git/Mailman/database, branch master

Bite the bullet: rename the Mailman package to mailman.

2008-02-27T06:26:18Z

Rework the basic infrastructure for qrunner process control. Split out the

2008-02-25T05:24:03Z

functionality of mailmanctl into a separate master watcher script. mailmanctl has not yet been updated but that'll happen next. Fix DELIVERY_MODULE to name a handler instead of a module. Change make_instance to use pkg_resources instead of module.__file__. Change the qrunner and master processes coordination so that the qrunners are not restarted on SIGINT, because otherwise C-c just doesn't work. Now SIGUSR1 is how we'll implement 'mailman restart'. Add a database commit so that initializing the schema doesn't lock the sqlite database. Also, don't try to initialize the schema if the tables already exist. Use some sqlite magic to do this test. Move mailman.cfg.in into a new package Mailman/extras inside the tree. Also, MAILMAN_UID and MAILMAN_GID should be integers not strings. Convert the command runner to use an IHandler instance instead of handler module. Similarly for the outgoing runner, DELIVERY_MODULE now names an IHandler instance instead of a handler module.

The start of a cleaning up of Errors.py. Eventually, I want to get rid of

2008-02-20T02:22:20Z

this module, in favor of moving exceptions into the interface modules that they are appropriate for. For now, this is just the low-hanging fruit. Along the way, clean up by reSTifying some interfaces and implementations.

Tweak copyright years.

2008-02-08T04:01:48Z

Convert IncomingRunner to use the new chains disposition architecture. move

2008-02-02T16:18:22Z

the big explanatory text at the beginning of incoming.py to a doctest called OVERVIEW.tt (which doesn't actually contain any tests yet -- it's documentation though). Added a doctest for the incoming runner, though this will be fleshed out in more detail next. Mailman.Post renamed to Mailman.inject, and simplified. We don't need its command line script behavior because that is now handled by bin/inject. Add a 'start_chain' attribute to mailing lists. This names the chain that processing of messages for that list begins with. We were inconsistent in the use of the 'no reply' address attribute. It's now always 'no_reply_address'. Update the smtplistener helper with lessons learned about how to suppress bogus asyncore error messages. Also, switch to using a maildir mailbox instead of an mbox mailbox.

SpamDetect is gone, so the chains/rules implementation experiment is deemed a

2008-02-02T03:21:05Z

success and will now be merged into the trunk. Move the Truth rule into the built-in rules package in a separate module, and add a test. Modify IChainLink so that the rule and chain attributes are not names but indeed the actual ILink or IChain object directly. Update the chains.process() function accordingly. Remove the IChain.get_rule() method. Don't derive BuiltInChain from Chain and don't make it an IMutableChain. It's now just an IChain, and is implemented concretely. Refactor the HeaderMatchChain and friends so that it can be used with both the global HEADER_MATCHES variable and the list-specific header_matches variable, which has exactly the same semantics. Oh yeah, get rid of the list's header_filter_rules attribute and replace it with header_matches so that the semantics match, it's easy to explain, and it's all nice and clean.

More updates to rules and chains, especially by adding additional tests.

2008-01-21T05:26:55Z

- Remove the Emergency.py handler. This is now covered by the emergency.py rule. - Remove the hold.txt doctest. This is now covered by the chains.txt and emergency.txt doctests. - Added a chains.txt doctest. - Extend the pending db to handle list values, although only lists that contain nothing but strings. - Rename the 'adminapproved' message metadata key (for the emergency rule) to 'moderator_approved'. Also, backport a change to the Decorate handler where message metadata on the 'decoration-data' key ca be used in the header and footer.

Use a more efficient way of deleting rows from a table, which coincidentally

2008-01-02T21:02:19Z

works around the storm cache bug #178546.

Add three new rules and their associated doctests.

2007-12-30T07:47:45Z

- A rule that checks to see if the sender is a moderated member. - A rule that checks to see if the sender is a non-member. - A rule that checks to see if the message has no (or an empty) Subject. Give IMembers (and the associated database implementation) an `is_moderated` flag.

Add .get() to our Message subclass, which ensures that returned

2007-12-11T04:00:14Z

values are unicodes if they come from the base class as a string. Get rid of the 'global id'. Now use just Message-ID. Rename X-List-ID-Hash to X-Message-ID-Hash. Do not take Date header into account when calculating this hash. Because of the above change, the assumption is that there will be no Message-ID collisions. Therefore, get rid of IMessageStore .get_message(), .get_messages_by_message_id() and .get_messages_by_hash(). Instead, it's now .get_message_by_id() and .get_message_by_hash() both of which return the message object or None. Message.hash -> Message.message_id_hash When storing a message in the message store, the final path component has the entire hash, not just the leftover parts after directory prefix splitting. MessageStore.delete_message() deletes the file too. Doctests clean up message store messages though the message store instead of directly off the filesystem.