Derek Lambert writes:
For new messages does mailman generate unique message IDs for each recipient?
Mailman in principle doesn't generate message IDs at all. That's the job of originating clients or their MTAs.
Mailman is what is called an "intermediary": it may do useful transformations on a message (such as adding RFC 2369 header fields and the List-Id field, as well as headers, footers, and subject decorations, removing banned media types, and even changing the format of text parts (such as removing HTML parts from multipart/alternative messages, or translating HTML to plain text by formatting with a browser such as Lynx), but in some sense the content of the message remains faithful to the author's intent. In such cases, the message is considered to be the same for the purposes of assigning Message-Id, so Mailman doesn't touch it.
In the case that Mailman receives a message that has no Message-Id, it may be a good idea to generate one. (I don't think this is implemented. It has its problems as well as its benefits.) Otherwise, no, Mailman leaves the existing Message-Id alone.
E.x. if multiple lists (children) are members of another list (parent), will the children lists all be able to archive the same message sent by parent?
Yes. I'm not sure whether the message body is stored multiple times or only once[1], but the primary key for archived message is effectively the list id plus the message-id.
Steve
Footnotes: [1] Probably multiple times, once for each list that has archiving turned on. Ensuring that any relevant decorations (headers, footers, Subject tags) are consistent would require pretty fragile analysis of the messages and complicated storage and indicies.