Follow kitty's Kitty graphics image lifetime rules #98

Merged

natsukium merged 4 commits from fix/image-lifetime into main 2026-10-12 00:39:59 +09:00

Conversation 0 Commits 4 Files changed 27 +685 -456

natsukium commented 2026-10-11 14:07:19 +09:00

Owner

Copy link

Part of #84 (image lifetime). Three commits:

1. Screen clears follow kitty's grman_clear: ED 2 / DECSED 2 / RIS drop every on-screen placement (C=1 included), ED 3 also scrollback ones, then every unreferenced image is freed. ED 0/1, EL and DECSTR no longer touch images. Stashed primary-screen placements count as references (kitty keeps that screen's images in its own store), which also fixes d=A on the alternate screen.
2. An image sent without i=/I= is freed once a delete or a scroll leaves it unplaced.
3. Byte-cap eviction follows apply_storage_quota: unreferenced images first, then least recently transmitted/placed.

Checked side by side with kitty: an EL redraw beside an image keeps it, ED 2 clears everything, and re-placing a cleared image answers ENOENT in both.

Part of #84 (image lifetime). Three commits: 1. Screen clears follow kitty's `grman_clear`: ED 2 / DECSED 2 / RIS drop every on-screen placement (`C=1` included), ED 3 also scrollback ones, then every unreferenced image is freed. ED 0/1, EL and DECSTR no longer touch images. Stashed primary-screen placements count as references (kitty keeps that screen's images in its own store), which also fixes `d=A` on the alternate screen. 2. An image sent without `i=`/`I=` is freed once a delete or a scroll leaves it unplaced. 3. Byte-cap eviction follows `apply_storage_quota`: unreferenced images first, then least recently transmitted/placed. Checked side by side with kitty: an EL redraw beside an image keeps it, ED 2 clears everything, and re-placing a cleared image answers `ENOENT` in both.

natsukium added 3 commits 2026-10-11 14:07:19 +09:00

grid, daemon: clear Kitty graphics images on the erases kitty clears on ... 669e8a448a

felis dropped every non-C=1 placement a partial erase touched (ED 0,
ED 1, EL) and every placement on DECSTR, while ED 2 spared C=1
placements and freed no image. kitty does the opposite on each count:
screen_erase_in_display calls grman_clear only for ED 2 and ED 3 (and
their selective DECSED twins), the clear ignores C=1, and it then frees
every image left without a reference; screen_reset does the same for
RIS, and DECSTR touches no image. A TUI that redraws one line beside an
image lost the image, and transmit-only images piled up after `clear`
until the byte cap evicted them.

The grid now reports a clear's scope instead of a row range, and the
daemon frees unreferenced images after it. The primary screen's stashed
placements count as references: kitty keeps that screen's images in a
store of its own, so vim clearing the alternate screen must not free
them. d=A shares the same path, which also closes that hole for it.

Assisted-by: Claude Code

grid, daemon: free an id-less Kitty graphics image with its last placement ... 6aa0db8133

An image transmitted without i= or I= can never be placed again, yet
felis kept it after its placements were deleted or scrolled away, so
yazi's anonymous previews held their share of the byte cap until
eviction pushed out images a producer could still use.

kitty frees such an image as soon as nothing references it: filter_refs
(every delete) and modify_refs (every scroll) both drop images with no
refs and no client id. The store now records which entries were sent
without an id, and the daemon sweeps those after each delete and each
scroll into history. The primary screen's stashed placements count as
references, as in the screen-clear path.

Assisted-by: Claude Code

grid, daemon: evict Kitty graphics images by kitty's storage quota rule ... 3f2e725cca

felis evicted strictly in transmit order, placed or not, so a still
image placed a moment ago could go before a transmit-only image nobody
displays. kitty's apply_storage_quota frees every unreferenced image
first and only then the oldest by atime, which it bumps on transmit and
on every put.

The store now keeps a logical clock and stamps an entry on insert and
on each real or virtual placement; the daemon trims unreferenced images,
then evicts the smallest stamp. Images the stashed primary screen shows
are left alone in both phases: kitty keeps each screen's images in its
own store, so a full-screen program cannot evict the shell screen's.

Assisted-by: Claude Code

natsukium force-pushed fix/image-lifetime from 3f2e725cca to 2ebb023c02 2026-10-11 14:14:27 +09:00 Compare

natsukium force-pushed fix/image-lifetime from 2ebb023c02 to 18e65c5197 2026-10-11 23:30:54 +09:00 Compare

natsukium force-pushed fix/image-lifetime from 18e65c5197 to d64070d119 2026-10-12 00:03:20 +09:00 Compare

natsukium force-pushed fix/image-lifetime from d64070d119 to ce639a5f8f 2026-10-12 00:07:15 +09:00 Compare

natsukium force-pushed fix/image-lifetime from ce639a5f8f to c5267686f4 2026-10-12 00:11:45 +09:00 Compare

natsukium merged commit 1190d1492e into main 2026-10-12 00:39:59 +09:00

natsukium deleted branch fix/image-lifetime 2026-10-12 00:39:59 +09:00

natsukium referenced this pull request from a commit 2026-10-12 00:39:59 +09:00

Merge pull request 'Follow kitty's Kitty graphics image lifetime rules' (#98) from fix/image-lifetime into main

natsukium referenced this pull request 2026-10-12 01:32:09 +09:00

graphics: smaller Kitty graphics differences from kitty to decide on #84

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!98

No description provided.