Stitching together serieses

Stitching together serieses

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 489 bytes --]

Hi,

When users send a patch series in a way that doesn't thread such as:

  https://lore.kernel.org/r/20260323041505.2088-1-niranjan.hy@ti.com
  https://lore.kernel.org/r/20260323041545.2101-1-niranjan.hy@ti.com
  https://lore.kernel.org/r/20260323041624.2120-1-niranjan.hy@ti.com
  https://lore.kernel.org/r/20260323041641.2133-1-niranjan.hy@ti.com

it would be useful if b4 review had a way for users to tell it that
actually this is a series and please treat it as such.

Thanks,
Mark

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: Stitching together serieses

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 949 bytes --]

On Mon, Mar 23, 2026 at 02:07:13PM +0000, Mark Brown wrote:
> Hi,
> 
> When users send a patch series in a way that doesn't thread such as:
> 
>   https://lore.kernel.org/r/20260323041505.2088-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041545.2101-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041624.2120-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041641.2133-1-niranjan.hy@ti.com
> 
> it would be useful if b4 review had a way for users to tell it that
> actually this is a series and please treat it as such.

Not just useful actually - at the minute I get:

  Retrieving series: 20260323041545.2101-1-niranjan.hy@ti.com
  Looking up https://lore.kernel.org/all/20260323041545.2101-1-niranjan.hy@ti.com/
  Grabbing thread from lore.kernel.org/all/20260323041545.2101-1-niranjan.hy@ti.com/t.mbox.gz
  Could not find cover letter or first patch

trying to track the second patch.


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: Stitching together serieses

[Konstantin Ryabitsev]

On Mon, Mar 23, 2026 at 02:07:13PM +0000, Mark Brown wrote:
> When users send a patch series in a way that doesn't thread such as:
> 
>   https://lore.kernel.org/r/20260323041505.2088-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041545.2101-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041624.2120-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041641.2133-1-niranjan.hy@ti.com
> 
> it would be useful if b4 review had a way for users to tell it that
> actually this is a series and please treat it as such.

I'll see what I can do, but on the first approach this is proving fragile,
because it's breaking a lot of assumptions we currently have about a series,
such as ability to retrieve it as a single thread to check for updates, etc.

bugspray tag b4/email

-- 
KR

Re: Stitching together serieses

[B4 Bugbot]

Hello:

This conversation is now tracked by b4 bug tracker.
There is no need to do anything else, just keep talking.

--
Deet-doot-dot, I am a bot.
b4 bug tracker (bugspray 0.1-dev)

Re: Stitching together serieses

[Konstantin Ryabitsev]

On Mon, Mar 23, 2026 at 02:07:13PM +0000, Mark Brown wrote:
> When users send a patch series in a way that doesn't thread such as:
> 
>   https://lore.kernel.org/r/20260323041505.2088-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041545.2101-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041624.2120-1-niranjan.hy@ti.com
>   https://lore.kernel.org/r/20260323041641.2133-1-niranjan.hy@ti.com
> 
> it would be useful if b4 review had a way for users to tell it that
> actually this is a series and please treat it as such.

After several bad takes, I landed --rethread in master that should do the
right thing in Most Situations (and probably screw things up badly in all
other situations, tbh).

E.g. this works:

    b4 review track --rethread https://lore.kernel.org/r/20260323041505.2088-1-niranjan.hy@ti.com

It also works for b4 am/shazam.

Please try it out and let me know any bugs you find.

Regards,
-- 
KR

Re: Stitching together serieses

[B4 Bugbot]

Konstantin Ryabitsev writes in commit 1718708bd211406cc60f6343b559fe02ad2be2a4:

Add --rethread flag for stitching unthreaded patch series

When submitters send patches without proper threading, maintainers
have no way to process them as a series with b4. Add a --rethread
flag to retrieval commands (am, shazam, mbox) and review track that
fetches unrelated messages and reconstitutes them into a properly
threaded series.

With a single message ID, b4 auto-discovers the rest of the series
by querying lore for patches from the same author within a 30-minute
window, matching by [PATCH n/m] counters and revision. With multiple
message IDs, b4 uses them directly. Stdin input is supported via
--rethread -.

When no cover letter is found, the first patch becomes the thread
root — the same structure as a normal coverless series on lore.
This avoids synthetic message IDs and keeps all outgoing references
(thank-you notes, trailers, review replies) pointing at real
messages on real mailing lists.

For review tracking, an is_rethreaded flag on the series table
(schema v7) gates the retrieval path so only rethreaded series use
the multi-fetch pipeline. The flag is persisted in the tracking
commit JSON so it survives database rebuilds via rescan. Member
patch information is stored in a new series_patches table for every
tracked series, laying the groundwork for future cherry-pick-at-
track-time support.

Key changes:
- Extract parse_msgid() from get_msgid() for reuse
- Add LoreSeries static methods for series reconstitution:
  identify_cover_letter(), rewrite_subject_counter(),
  renumber_patches(), rethread_messages(), rethread_series()
- Add discover_rethread_series() for auto-discovery from a
  single seed message using lore search (dt: time range queries)
- Add fetch_rethread_messages() and retrieve_rethreaded_messages()
- Hook into retrieve_messages() and review track's cmd_track()
- New series_patches table with (change_id, revision, position) PK
- New is_rethreaded column on series table for retrieval routing
- Persist is-rethreaded in tracking commit JSON; read it back in
  rescan_branches() and create_review_branch()
- retrieve_series_messages() gateway for review TUI retrieval
- ReviewApp._fetch_rethreaded_threads() for follow-up loading
- 45 tests covering parsing, renumbering, rethreading, discovery
  filtering, and integration with LoreMailbox
- Document --rethread in mbox, am/shazam, review, and man page

Closes: https://msgid.link/8637ff61-4eca-42ac-96fb-d530498b3f82@sirena.org.uk # 4958fdf
Reported-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
Assisted-by: claude-opus-4-6

--
Deet-doot-dot, I am a bot.
b4 bug tracker (bugspray 0.1-dev)