natsukium/felis-web-gateway
Watch
1
Code Issues 1 Pull requests Projects Releases Packages Wiki Activity Actions

Follow felis to v0.1.0 #4

Merged
natsukium merged 2 commits from follow-felis-v0.1.0 into main 2026-09-26 20:53:18 +09:00
Conversation 0 Commits 2 Files changed 8 +47 -39
natsukium commented 2026-09-26 19:02:26 +09:00
Owner
Copy link

Moves the four felis crates from 6da5620d on the archived private repository to the v0.1.0 tag on https://github.com/felis-terminal/felis (the crates are not published to crates.io, so the tag is the stable name).

What broke

  • felis split each two-way IPC family into one wrapper per direction. The gateway now writes SessionToDaemonMsg (attach/theme/detach) and NotifyToDaemonMsg::Subscribe, and decodes NotifyToClientMsg::Event. Mechanical rename; no behavior change.
  • nix/gateway.nix outputHashes refreshed for the new git source.
  • README/docs/broker.md/Cargo comments: pinned rev → tag, 6da5620d → v0.1.0, felis link → GitHub.

The felis-json v1 frames the browser receives are unchanged (json_v1::encode decodes Session/Conn bodies as either direction's wrapper), so felis-web-component only needs the same pin bump.

Verification

  • cargo fmt --all --check, cargo clippy --all-targets -- -D warnings, cargo test (27 passed), cargo deny check, nix build .#gateway: all pass locally.
  • Live smoke against a v0.1.0 felis-daemon on an isolated socket: broker list / attach (8-digit prefix) / input / detach round-trip with minor: 0 and v1-enveloped grid frames; the one-session felis.v1 socket streams the rehydrate burst and echoes input.

Issue #1 (search/export) stays open: it is a feature restoration with a design choice, which this follow does not change.

Moves the four felis crates from `6da5620d` on the archived private repository to the `v0.1.0` tag on https://github.com/felis-terminal/felis (the crates are not published to crates.io, so the tag is the stable name). ## What broke - felis split each two-way IPC family into one wrapper per direction. The gateway now writes `SessionToDaemonMsg` (attach/theme/detach) and `NotifyToDaemonMsg::Subscribe`, and decodes `NotifyToClientMsg::Event`. Mechanical rename; no behavior change. - `nix/gateway.nix` `outputHashes` refreshed for the new git source. - README/`docs/broker.md`/Cargo comments: pinned `rev` → `tag`, `6da5620d` → `v0.1.0`, felis link → GitHub. The felis-json v1 frames the browser receives are unchanged (`json_v1::encode` decodes Session/Conn bodies as either direction's wrapper), so felis-web-component only needs the same pin bump. ## Verification - `cargo fmt --all --check`, `cargo clippy --all-targets -- -D warnings`, `cargo test` (27 passed), `cargo deny check`, `nix build .#gateway`: all pass locally. - Live smoke against a v0.1.0 `felis-daemon` on an isolated socket: broker `list` / `attach` (8-digit prefix) / `input` / `detach` round-trip with `minor: 0` and v1-enveloped grid frames; the one-session `felis.v1` socket streams the rehydrate burst and echoes input. Issue #1 (search/export) stays open: it is a feature restoration with a design choice, which this follow does not change.
chore: follow felis to v0.1.0
Some checks failed
ci / fmt / clippy / test / deny (pull_request) Has been cancelled
Details
dadcf2bb36
 
felis v0.1.0 is the first public release and the first one this relay
can pin by a stable name: the crates are not on crates.io, so the
dependency now points at the tag on the public GitHub repository
instead of a commit hash on the archived private one. The browser
halves decode the same frames, so felis-web-component moves to the
same tag.

The only API break between the previous pin and the release is the
split of the two-way IPC families into one wrapper per direction:
the gateway writes `SessionToDaemonMsg` (attach, theme, detach) and
`NotifyToDaemonMsg::Subscribe`, and reads `NotifyToClientMsg::Event`
back. The felis-json v1 shape the browser receives is unchanged: the
envelope, the kind tokens and every DTO are the ones the last follow
already emits, and `json_v1::encode` now decodes a Session or Conn
body as either direction's wrapper, so the relay needs no dispatch of
its own.

Checked live against a v0.1.0 daemon: the broker's list, attach,
input and detach round-trip, and a one-session socket streams the
rehydrate burst and echoes input.
ci: run on the x86_64-linux runner label
All checks were successful
ci / fmt / clippy / test / deny (pull_request) Successful in 1m14s
Details
953b62f3f8
 
The Forgejo runners were re-registered with labels named after Nix
systems (nix -> x86_64-linux, macos -> aarch64-darwin), so jobs asking
for the old labels sit in the queue with no runner to pick them up.
natsukium referenced this pull request from a commit 2026-09-26 20:53:20 +09:00
Follow felis to v0.1.0 (#4)
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
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-web-gateway!4
No description provided.