[Mailman-Developers] Message Handling Workflow

[Stephen J. Turnbull]

VuNghia Truong writes:

 > Could any one please help me to answers these questions:
 > - Is the handling workflow b.t version 2 and 3 the same or different?

It's different.  V2 has a single pipeline that both checks the mail
for conformance to list regulations and also cleans up the post
(removing forbidden content types) and adds list-specific content
(mail headers, leader and trailer parts).  V3 has two separate
pipelines, one for approving the post (using "chains" of "rules") and
one for preparting for distribution (using a "pipeline" of
"handlers").  Be carefule because the terminology resembles that used
in V2 but the architecture is definitely different.

 > - How the file system are organized, how these pieces of codes fit
 >   together?

The /var hierarchy for Mailman 3 is quite similar to that of Mailman
2.  I'm not sure what the installed code looks like; we currently
recommend running in a virtualenv, basically everything in the source
tree.

The source hierarchy is quite different.  There are altogether 4
separate source trees:

- Mailman 3 core: the post distribution and databases
- mailman.client: helper library for accessing database and
  configuration using the REST (HTTP) interface
- Postorius: the web admin interface (including user subscription
  options), which depends on mailman.client
- HyperKitty: the archive management and web access interface.

Note that Postorius and HyperKitty are designed to function as
independent modules, and are basically independent projects.