Follow kitty's placement identity and lifetime rules #87

Merged

natsukium merged 3 commits from fix/idless-placements into main 2026-10-10 11:47:16 +09:00

Conversation 0 Commits 3 Files changed 14 +358 -26

natsukium commented 2026-10-09 23:23:27 +09:00

Owner

Copy link

Closes #78. Closes #79.

The alternate screen is out of scope: felis shares one image store between screens where kitty keeps one per screen, which is #89. The placement table's size bound is #88.

daemon, grid: delete an image's placements when it is transmitted again. Re-transmitting under a live image id left its placements on screen showing the new pixels, with no ref pinning the image; kitty deletes them (free_refs_data). Re-placing under the same (i, p) also took a second ref while replacing the placement, pinning the image after it was deleted. Both are stated as one property over puts, deletes and re-transmits against a model, with every live placement holding exactly one ref.

grid, daemon: give each Kitty placement without p= a placement of its own. The table kept one id-less slot per image, so placing an image twice without p= moved the first copy; kitty replaces only under a non-zero p=.

• An id-less put (no p=, or p=0) gets an id counted down from u32::MAX, like an unnamed image's id, and is marked anonymous. The wire and client shadow key on (image, placement id) and need no change.
• No producer p= addresses an anonymous placement: named puts and deletes skip it, and a named put landing on its id first moves it to a fresh one at the end of the table (announced as a remove and a re-place, which the client appends).
• A delete with p=0 counts as one without p=; p= on an image sent without i=/I= is ignored; the reply echoes p= only when non-zero.
• The property gains id-less, p=0 and u32::MAX ids; each of the four fixes above was removed by hand and the property failed every time.

daemon: free an image on an uppercase delete only once it is unplaced. d=I with p= and the d=P/Q/X/Y/Z/C/R predicates freed the image even when other placements still used it, orphaning them; kitty's filter_refs frees only once no refs remain. The property gains d=I and tracks whether the image is still stored.

Verification

• Raw-escape repro in a felis window (headless sway screenshot): two a=p,i=1,c=10,r=5,C=1 at different cells draw two copies, as in kitty; before, one.
• The alt-screen property passes unchanged with id-less placements. the_mirror_keeps_the_daemons_placement_order ships every step to a client shadow and checks it holds the daemon's placements in the daemon's order.
• just check passes apart from six felis-shaping tests that fail on main too on this host (pinned font files missing).

Closes #78. Closes #79. The alternate screen is out of scope: felis shares one image store between screens where kitty keeps one per screen, which is #89. The placement table's size bound is #88. **daemon, grid: delete an image's placements when it is transmitted again.** Re-transmitting under a live image id left its placements on screen showing the new pixels, with no ref pinning the image; kitty deletes them (`free_refs_data`). Re-placing under the same `(i, p)` also took a second ref while replacing the placement, pinning the image after it was deleted. Both are stated as one property over puts, deletes and re-transmits against a model, with every live placement holding exactly one ref. **grid, daemon: give each Kitty placement without p= a placement of its own.** The table kept one id-less slot per image, so placing an image twice without `p=` moved the first copy; kitty replaces only under a non-zero `p=`. - An id-less put (no `p=`, or `p=0`) gets an id counted down from `u32::MAX`, like an unnamed image's id, and is marked anonymous. The wire and client shadow key on `(image, placement id)` and need no change. - No producer `p=` addresses an anonymous placement: named puts and deletes skip it, and a named put landing on its id first moves it to a fresh one at the end of the table (announced as a remove and a re-place, which the client appends). - A delete with `p=0` counts as one without `p=`; `p=` on an image sent without `i=`/`I=` is ignored; the reply echoes `p=` only when non-zero. - The property gains id-less, `p=0` and `u32::MAX` ids; each of the four fixes above was removed by hand and the property failed every time. **daemon: free an image on an uppercase delete only once it is unplaced.** `d=I` with `p=` and the `d=P/Q/X/Y/Z/C/R` predicates freed the image even when other placements still used it, orphaning them; kitty's `filter_refs` frees only once no refs remain. The property gains `d=I` and tracks whether the image is still stored. ## Verification - Raw-escape repro in a felis window (headless sway screenshot): two `a=p,i=1,c=10,r=5,C=1` at different cells draw two copies, as in kitty; before, one. - The alt-screen property passes unchanged with id-less placements. `the_mirror_keeps_the_daemons_placement_order` ships every step to a client shadow and checks it holds the daemon's placements in the daemon's order. - `just check` passes apart from six `felis-shaping` tests that fail on `main` too on this host (pinned font files missing).

natsukium added 1 commit 2026-10-09 23:23:27 +09:00

grid, daemon: give each Kitty placement without p= a placement of its own ... 3eaf4ad68b

The placement table keyed an id-less put as the image's single default
slot, so placing one image twice without p= moved the first copy to the
second position. kitty replaces only under a non-zero placement id:
handle_put_command reuses a ref only when p is set and the image has a
client id, and the spec says id-less puts of one image "result in
multiple placements".

An id-less put now gets an id from the placement table, counted down
from u32::MAX and skipping the image's live ids, the way an unnamed
image gets its image id. The wire and the client shadow already key on
(image, placement id), so they need no change and tell the copies
apart. p= on an image sent without i= or I= is ignored, as in kitty, and
the reply echoes p= only when it is non-zero, which the property over
mixed id-less, p=0 and named puts turned up.

The alt-screen property now places with p=1: it re-transmits under one
image id, and with id-less puts accumulating, the store's refcount reset
on re-transmit (which leaves the old placements behind) would break its
refcount invariant. That reset is a separate bug.

Assisted-by: Claude Code

natsukium force-pushed fix/idless-placements from 3eaf4ad68b to 70f4138d37 2026-10-09 23:39:31 +09:00 Compare

natsukium changed title from Give each Kitty placement without p= a placement of its own to Follow kitty's placement identity and lifetime rules 2026-10-09 23:56:07 +09:00

natsukium force-pushed fix/idless-placements from 70f4138d37 to b022cb0eea 2026-10-09 23:56:08 +09:00 Compare

natsukium force-pushed fix/idless-placements from b022cb0eea to 34f217e80c 2026-10-10 04:07:38 +09:00 Compare

natsukium force-pushed fix/idless-placements from 34f217e80c to 13cf708741 2026-10-10 04:19:08 +09:00 Compare

natsukium merged commit e12311c207 into main 2026-10-10 11:47:16 +09:00

natsukium deleted branch fix/idless-placements 2026-10-10 11:47:16 +09:00

natsukium referenced this pull request from a commit 2026-10-10 11:47:18 +09:00

Merge pull request 'Follow kitty's placement identity and lifetime rules' (#87) from fix/idless-placements into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

Area/ci

CI workflows (.forgejo/workflows/)

  

Area/cli

felis-cli crate, the felis front-door binary

  

Area/client

felis-client crate, the winit GUI frontend

  

Area/client-core

felis-client-core crate, shadow screen, config, connect_or_spawn_daemon

  

Area/daemon

felis-daemon crate, session pool, IPC server

  

Area/deps

Dependency bumps

  

Area/docs

Documentation tree when no narrower area fits

  

Area/grid

felis-grid crate, cells, scrollback, image store

  

Area/manual

User-facing docs: tutorials, how-to, user-facing reference pages

  

Area/nix

Nix flake, home-manager module, package definitions

  

Area/protocol

felis-protocol crate, felis.proto, frame layout, version constants

  

Area/pty

felis-pty crate, PTY backends

  

Area/render

felis-render-wgpu crate, GPU pipeline, atlas

  

Area/shaping

felis-shaping crate, swash shaper, font fallback

  

Area/skills

Agent skills (skills/felis, .agents/skills/)

  

Area/transport

felis-transport crate, framing, socket/pipe, SSH carrier

  

Area/vt

felis-vt crate, VT parser, dispatch

  

Compat/Breaking

Breaking change that won't be backward compatible

  

Kind/Bug

Something is not working

  

Kind/Documentation

Documentation changes

  

Kind/Enhancement

Improve existing functionality

  

Kind/Feature

New functionality

  

Kind/Performance

Throughput, latency, or memory cost

  

Kind/Security

This is security issue

  

Kind/Testing

Issue or pull request related to testing

  

perf-regression-accepted

Passes the instruction-count bench gate for an intended cost; state why in a PR comment

  

Reviewed

Confirmed

Issue has been confirmed

  

Reviewed

Duplicate

This issue or pull request already exists

  

Reviewed

Invalid

Invalid issue

  

Reviewed

Won't Fix

This issue won't be fixed

  

Status

Blocked

Something is blocking this issue or pull request

  

Status

Need More Info

Feedback is required to reproduce issue or to continue work

  

Status

Upstream

Waiting on a change in an upstream project

No labels

Area/ci

Area/cli

Area/client

Area/client-core

Area/daemon

Area/deps

Area/docs

Area/grid

Area/manual

Area/nix

Area/protocol

Area/pty

Area/render

Area/shaping

Area/skills

Area/transport

Area/vt

Compat/Breaking

Kind/Bug

Kind/Documentation

Kind/Enhancement

Kind/Feature

Kind/Performance

Kind/Security

Kind/Testing

perf-regression-accepted

Reviewed

Confirmed

Reviewed

Duplicate

Reviewed

Invalid

Reviewed

Won't Fix

Status

Blocked

Status

Need More Info

Status

Upstream

Milestone

Clear milestone

No items

No milestone

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

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

felis-terminal/felis!87

No description provided.