Concepts

Series and families

A series is one posting: a cover letter (or a lone first patch) and the patches threaded under it. A family is every version of one series, the nearest email has to a pull request. Versions are grouped by b4’s change-id (from the cover letter’s trailer or a b4 message-id), otherwise by the series title.

A family is judged on its latest version’s patches, each as far as any version of it got. A patch is followed from version to version by title, then by an identical diff (a retitled patch), then by a similar title. A patch a later version dropped still counts if it was applied somewhere: dropping a patch because it went in is normal. One dropped and never applied was abandoned and does not hold the family back.

Who a series belongs to

A series is yours if you posted it or wrote it.

• The sender is X-Original-From: when a relay or list rewrote From: (b4’s web relay sends as “X via B4 Relay <devnull+…>”), otherwise From:.

• The author is the in-body From: at the top of a patch (what git am records), otherwise the sender.

So a patch you resend for someone else is on your list (“written by …; posted by you” in show), and so is your patch that a maintainer resends.

States

A series is as far along as its least advanced patch, and every state carries the message or commit that shows it.

state	evidence
merged	every patch found in a mainline tree
in-next	every patch found in a next tree (or further)
queued	every patch found in a maintainer tree (or further)
applied	a reply from someone else says “Applied …”, “Queued …”, “Thanks, applied”, “This was merged into …”, or the patchwork-bot “This series was applied to …”, covering every patch. A b4-ty [2/3] … commit: list names the patches; a bare reply to the cover covers all of them.
partially-applied	some patches got at least as far as applied, others did not (a b4-ty “(subset)”, or “…, others to follow”)
superseded	a later version, or a RESEND, of the same family exists
commented	a person’s latest direct reply to a patch (or the cover) wrote prose below quoted text and gave no tag
reviewed	a Reviewed-by: or Acked-by:, or a project’s approval (see Other projects)
discussed	at least one reply from a person
bot-only	replies only from bots
no-reply	the thread was read and nobody else wrote
unfetched	no replies known, but the thread has never been read
closed	you closed it (patchtrack close)

applied comes from the mail alone. When trees are configured and none of them has the patches, the evidence says so: "Applied to the for-next branch…"; not in foo as of 2026-10-26 21:12. That is the usual state between a maintainer’s reply and their push.

commented is the nearest email has to GitHub’s “changes requested”. Mail cannot say whether the author is being asked for something, so the state names what was done: someone commented inline. Only direct replies to a posting count; replies to replies are mostly conversation. A later direct reply carrying a tag settles that patch.

As of

Every state is as of the time its thread was last read. “No reply” is a timestamp, not a fact. A thread read only from your own mail says “from your mailbox only”, because local mail misses replies that dropped you from Cc.

unanswered marks a series where a person wrote after you last did. It is a fact, not a judgement: a bare Reviewed-by counts too. nudge is stricter about what needs a reply.