b4 review handling when base can't be guessed

b4 review handling when base can't be guessed

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 371 bytes --]

Hi,

In b4 review tui if the base of a series can't be guessed then when
trying to review you get stopped on a screen saying "Attestation check"
and in the dialog "Base: not specified (a/b blobs match HEAD)".  It
would be really helpful to have a flow for specifying what HEAD to try
here other than quitting and doing a git checkout (which I hope works?).

Thanks,
Mark

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

Re: b4 review handling when base can't be guessed

[Konstantin Ryabitsev]

On Tue, Mar 10, 2026 at 12:49:41PM +0000, Mark Brown wrote:
> In b4 review tui if the base of a series can't be guessed then when
> trying to review you get stopped on a screen saying "Attestation check"
> and in the dialog "Base: not specified (a/b blobs match HEAD)".  It
> would be really helpful to have a flow for specifying what HEAD to try
> here other than quitting and doing a git checkout (which I hope works?).

It shouldn't stop you there -- you can still "y" confirm and try to apply it
to a best-guessed tree.

However, let me play with it and see if we can also present an override for
which base we use in that screen.

bugspray tag b4/review
-- 
KR

Re: b4 review handling when base can't be guessed

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 825 bytes --]

On Tue, Mar 10, 2026 at 08:59:37AM -0400, Konstantin Ryabitsev wrote:
> On Tue, Mar 10, 2026 at 12:49:41PM +0000, Mark Brown wrote:
> > In b4 review tui if the base of a series can't be guessed then when
> > trying to review you get stopped on a screen saying "Attestation check"
> > and in the dialog "Base: not specified (a/b blobs match HEAD)".  It
> > would be really helpful to have a flow for specifying what HEAD to try
> > here other than quitting and doing a git checkout (which I hope works?).

> It shouldn't stop you there -- you can still "y" confirm and try to apply it
> to a best-guessed tree.

That didn't appear to work since it looks like it uses the same logic to
try to apply the patch:

| Guessing base commit...
| Base: failed to guess (None of the 1 blob(s) matched)
| Could not determine base commit

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

Re: b4 review handling when base can't be guessed

[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: b4 review handling when base can't be guessed

[B4 Bugbot]

Konstantin Ryabitsev writes in commit 9189f0b31a20311748625d0f2f88afbcbc41f4e4:

Add base selection, attestation display, and take confirmation to review TUI

When the base of a series can't be guessed, the review TUI previously
offered no way to specify an alternative base without quitting and
manually checking out a different branch. This was reported by Mark
Brown (bug 38491be).

Improve the review tracking TUI with several interrelated features that
give maintainers better control and visibility during patch review.

Base commit selection: add BaseSelectionScreen modal that lets
maintainers select or override the base commit when checking out a
series for review. The screen tests patch applicability at the chosen
base and shows the resolved SHA on success. A reusable WorkerScreen
modal provides a loading indicator for running callables in worker
threads.

Attestation display: add a checkmark column (A) to the series listing
that shows green when all patches have valid DKIM/patatt signatures.
The details pane shows per-attestor breakdown after the From line.
Schema v5 adds an attestation column to the tracking database, and
check_series_attestation() checks signatures during update and
checkout, storing per-attestor results (signed/nokey/badsig).

Take confirmation: add TakeConfirmScreen as a final step in the take
flow that test-applies patches before executing. This gives maintainers
a chance to back out if the series will not apply cleanly to the target
branch. Ctrl-y is now the confirmation key consistently across all take
flow screens (TakeScreen, CherryPickScreen, TakeConfirmScreen).

Keybinding changes: split the update keybinding so that u updates only
the selected series while U updates all tracked series, with proper
footer grouping. Cursor position is preserved after update operations.

Key changes:
- BaseSelectionScreen with test-apply and resolved SHA display
- WorkerScreen reusable modal for worker thread callables
- Schema v5 migration for attestation column
- check_series_attestation() and _format_attestation() helpers
- TakeConfirmScreen with pre-take applicability check
- u/U keybinding split with cursor preservation
- Documentation updated for all new features
- 27 new tests for attestation and formatting

Reported-by: Mark Brown <broonie@kernel.org>
Closes: https://msgid.link/092a612b-4fa1-4fa7-a872-b3564fcb3749@sirena.org.uk # 38491be
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)