protocol, transport, daemon: give the preface replies and key ids narrow types #472

Merged

natsukium merged 5 commits from cleanup/preface into main 2026-09-26 03:17:25 +09:00

Conversation 1 Commits 5 Files changed 5 +77 -57

natsukium commented 2026-09-26 01:32:12 +09:00

Owner

Copy link

Why

Pre-release cleanup. The daemon built its preface reply with DaemonPreface::select() / refuse(), whose signatures return the full three-arm wire enum although each only ever builds one arm, so exchange_bootstrap destructured the value it had just built behind two unreachable!s. named_key_to_wire had the same shape: Option whose None meant "an F key", re-matched by its caller behind a third unreachable!.

What

• felis-protocol: DaemonAccept { major, minor } with select(ClientPreface) -> Option<DaemonAccept> and DaemonRefuse { min_major, max_major } with CURRENT, each Into<DaemonPreface>. The decode-side DaemonPreface is unchanged.
• felis-transport: write_daemon_preface takes impl Into<DaemonPreface>.
• felis-daemon: exchange_bootstrap reads the narrow values; both unreachable!s are gone.
• named_key_to_wire returns the v1::key_id::Key arm directly; the caller's unreachable! is gone.
• New byte assertion pins DaemonRefuse::CURRENT's encoding.

Wire unchanged: preface bytes and key encoding are identical (just proto-compat green).

Verification

• cargo nextest run -p felis-protocol -p felis-daemon -p felis-transport -p felis-client-core --all-features (1476 passed), clippy -D warnings incl. felis-cli, just schema no diff, just prose-check: green.
• pi review (sol): PASS, no findings.

## Why Pre-release cleanup. The daemon built its preface reply with `DaemonPreface::select()` / `refuse()`, whose signatures return the full three-arm wire enum although each only ever builds one arm, so `exchange_bootstrap` destructured the value it had just built behind two `unreachable!`s. `named_key_to_wire` had the same shape: `Option` whose `None` meant "an F key", re-matched by its caller behind a third `unreachable!`. ## What - felis-protocol: `DaemonAccept { major, minor }` with `select(ClientPreface) -> Option<DaemonAccept>` and `DaemonRefuse { min_major, max_major }` with `CURRENT`, each `Into<DaemonPreface>`. The decode-side `DaemonPreface` is unchanged. - felis-transport: `write_daemon_preface` takes `impl Into<DaemonPreface>`. - felis-daemon: `exchange_bootstrap` reads the narrow values; both `unreachable!`s are gone. - `named_key_to_wire` returns the `v1::key_id::Key` arm directly; the caller's `unreachable!` is gone. - New byte assertion pins `DaemonRefuse::CURRENT`'s encoding. Wire unchanged: preface bytes and key encoding are identical (`just proto-compat` green). ## Verification - `cargo nextest run -p felis-protocol -p felis-daemon -p felis-transport -p felis-client-core --all-features` (1476 passed), clippy `-D warnings` incl. felis-cli, `just schema` no diff, `just prose-check`: green. - pi review (sol): PASS, no findings.

natsukium added 2 commits 2026-09-26 01:32:12 +09:00

protocol, transport, daemon: give the preface replies narrow types ... 39b74e320a

`DaemonPreface::select` returned `Option<DaemonPreface>` and `refuse`
returned `DaemonPreface`, yet each only ever built one variant. The
daemon's bootstrap then destructured the reply back into that variant
behind two `unreachable!`s, a panic path the type system could have
ruled out.

`DaemonAccept::select` and `DaemonRefuse::CURRENT` now carry exactly the
reply they build, and `write_daemon_preface` takes `impl
Into<DaemonPreface>` so the widening happens once at the write. The
decoded three-arm `DaemonPreface` is unchanged, as are the preface bytes;
a new assertion pins the bytes of this build's refusal.

Assisted-by: Claude Code

protocol: return the key id arm from named_key_to_wire ... 8a3317c4dc

`named_key_to_wire` returned `None` for F keys alone, so its only caller
matched on the key a second time and kept an `unreachable!` for the
non-F `None` case the first match already excluded. Returning the
`KeyId` oneof arm directly makes the split total: F keys go to
`function`, every other named key to the `NamedKey` enum. The encoded
bytes are unchanged.

Assisted-by: Claude Code

natsukium added 1 commit 2026-09-26 02:29:41 +09:00

Merge remote-tracking branch 'origin/main' into cleanup/preface 54fb83088e

natsukium added 1 commit 2026-09-26 02:42:46 +09:00

ci: rerun the bench gate after a load-noise failure ... 31f55e6793

The Criterion gate flagged daemon_side and multi_cycle benches the
preface change does not reach while the runner host was at load ~10.

Assisted-by: Claude Code

natsukium added 1 commit 2026-09-26 02:57:34 +09:00

Merge remote-tracking branch 'origin/main' into cleanup/preface 60fe12c9e7

natsukium merged commit 291d6a9fb4 into main 2026-09-26 03:17:25 +09:00

natsukium deleted branch cleanup/preface 2026-09-26 03:17:25 +09:00

natsukium referenced this pull request from a commit 2026-09-26 03:17:26 +09:00

protocol, transport, daemon: give the preface replies and key ids narrow types (#472)

natsukium commented 2026-09-26 03:17:36 +09:00

Author

Owner

Copy link

Merged with the Criterion gate red. It flagged daemon_side/plaintext_scroll (+39 to +45%) three times, in runs that overlapped other CI jobs on the same host. A local A/B on a quiet host, with a separate target dir for each side, found no difference:

round	main (a15562e8)	this branch
1	59.8 µs	61.2 µs
2	60.1 µs	58.7 µs
3	59.4 µs	60.7 µs

daemon_side/shell_prompt swings between about 31 µs and 64 µs within one checkout, on both sides. The daemon_side group is bimodal on this runner, and the branch only touches the preface handshake and key-id encoding, which these benches never reach.

Merged with the Criterion gate red. It flagged `daemon_side/plaintext_scroll` (+39 to +45%) three times, in runs that overlapped other CI jobs on the same host. A local A/B on a quiet host, with a separate target dir for each side, found no difference: | round | main (a15562e8) | this branch | |---|---|---| | 1 | 59.8 µs | 61.2 µs | | 2 | 60.1 µs | 58.7 µs | | 3 | 59.4 µs | 60.7 µs | `daemon_side/shell_prompt` swings between about 31 µs and 64 µs within one checkout, on both sides. The daemon_side group is bimodal on this runner, and the branch only touches the preface handshake and key-id encoding, which these benches never reach.

Commenting is not possible because the repository is archived.

Reviewers

No reviewers

Labels

Clear labels   

priority/P0

Must be resolved before the v0.1.0 tag

  

priority/P1

Must be settled before the first public contract freezes

  

priority/P2

Explicitly deferred until after v0.1.0

  

release/v0.1.0

Work tracked for the first tagged release

  

status/blocked

Cannot start now: waits on an upstream release, another issue, or a recorded revisit trigger

  

status/planned

Triaged with a source-grounded plan and a passed pi review; ready to implement

  

type/bug

Shipped behavior that contradicts the docs or the code's own intent; needs a product code change

  

type/design

An open scope or trade-off question; needs a recorded decision in the owning explanation doc before code

  

type/test-gap

Documented or reviewed behavior that no test pins; usually split from a review finding, needs no product code change

  

type/tracker

Parent issue that tracks child work

No labels

priority/P0

priority/P1

priority/P2

release/v0.1.0

status/blocked

status/planned

type/bug

type/design

type/test-gap

type/tracker

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

natsukium/felis!472

No description provided.