feat(saves): market-entrance, child Link in Castle Town Market's gate, by day #1723

Merged
HumanBean merged 1 commit from HumanBean/Longshot:feat/save-market-entrance into main 2026-11-21 05:34:30 +00:00
Owner

Type: feat · Risk: low · Tested on: CachyOS Linux, RTX 3090 (NVIDIA 620.70.06) · Needs: none

TL;DR

The Test Farm can only start a run from Link's house in Kokiri Forest, so a test that walks into Castle Town's market had to teleport and walk there first. This adds a repository save, market-entrance: child Link in the Market's gate, the town ahead of him, by day, its first-visit scene already seen. It was made from a fresh profile and scrubbed like kokiri-start; it passes check-repo-saves and loads there in the game.

Breakdown

  • Who it helps: the Test Farm and anyone testing a change in Castle Town's market, which is one of the game's busiest places.
  • What was wrong: the only repository save is in Kokiri Forest; a teleport reaches the Market, but a run that starts with a load there is the same on every PC and skips the first-visit scene.
  • What changes: a new save, assets/saves/market-entrance, and its row in the saves' README.
  • What does not change: the game, the scripts, kokiri-start; nothing for a player.
  • How it was checked: check-repo-saves passes; launched with --repo-save market-entrance, it loads in the Market's gate (state play, child, no event).
  • What is still open: nothing.
  • Risk: low: two save files and README text, read only by launch --repo-save and check-repo-saves.

Why

The Test Farm's planner asked for a town-entry save for tests of the render targets in the Market (#1255, #1256): kokiri-start is the only repository save, so a run had to teleport there itself, and the first walk in plays the Market's first-visit scene. With this save a run starts in the Market's gate, by day, the scene already seen, in the same place on every PC. The tracker has no issue for it (searched: "repo save", "assets/saves").

What changed

A new repository save, market-entrance, made fresh from a new profile and scrubbed with longshot_saves.py scrub, and its row in the saves' README.

  • assets/saves/market-entrance/slot_00/: progress.sav (the game's save, unchanged) and caption.sav (its picture mid-grey, its date 2026-01-01 00:00). No account.sav or option.sav.
  • assets/saves/market-entrance/README.md: where the save stands (its first line) and how it was made.
  • assets/saves/README.md: the save's row in the table.

How it was made, on this PC (CachyOS Linux, RTX 3090, NVIDIA 620.70.06; Longshot v3.0.0-rc.7-111-gda91e7702, main da91e770): a new profile with only kokiri-start in it (longshot_ctl.py launch <game> --profile test --repo-save kokiri-start --headless, an empty options file so every setting is at its default and the Autosave row On), Continue on the title, then through the control bridge teleport Market_Child (Link arrives in the Market's gate facing the town; that arrival writes no save: the save trace logged the requests and no slot changed), the left stick held back for 240 frames (out through the gate into Hyrule Field: an autosave, slot_01, the castle-gate save of its own PR), then back for 120 frames (through the gate into the Market again). The Market's first-visit scene played by itself, and the game's own autosave written after it (slot_02: scene Market, map Market/Market_Child, 3 hearts, by day) is this save; loading it puts Link where the teleport does, the scene already seen. Nothing else was done in the game; the story is a new game's, as in kokiri-start. Scrubbed on this PC, so the original picture never left it:

caption.sav: the save date (month at 280): 11 -> 1
caption.sav: the save date (day at 312): 19 -> 1
caption.sav: the save date (hour at 328): 2 -> 0
caption.sav: the save date (minute at 304): 16 -> 0
caption.sav: the save date (second at 288): 39 -> 0
caption.sav: the picture at 1396 (JPEG 543x306, 38241 bytes) replaced by a mid-grey JPEG 543x306 of 3345 bytes
caption.sav: 39637 -> 4741 bytes; the rest of its header kept as it was (scene Market, map Market/Market_Child)
progress.sav: copied unchanged

The save holds no game data: check-repo-saves passes, and progress.sav's readable strings are only the game's own names (Market_Child, Goto_HyruleCastle, Item_Wallet_Small and the like), with no paths or names from the PC.

Measurements

Nothing to measure: this adds a save and a README row, no code; it is checked by check-repo-saves, the saves' tests and a load in the game (Testing).

Flags and settings

None

Testing

Tested:

  • python -I scripts/longshot_saves.py check-repo-saves: passes.
  • tests/test_longshot_saves.py (77 tests, OK, 1 skipped) and tests/test_longshot_ctl.py (51 tests, OK), on Linux.
  • python -I -B tests/run_all.py on this PC (Linux, a jj workspace of this branch on main @ 9b82abf6, before the rebase onto 99a7e249, which touched nothing here): 692 passed, 37 failed, 97 skipped. 36 of the failures also fail on main @ 9b82abf6 in a clean workspace (run_all.py --exact on them: 36 of 36 fail; they need tools or files that workspace lacks): test_bug_report, test_bug_report_defaults, test_capture_read_once, test_crash_groups, test_crt_options, test_damage_debug_menu, test_dev_packs, test_display_startup_failure, test_dlss_feature_size, test_failed_compile_keeps_objects, test_guest_imported, test_hardware_report, test_high_fps_fixes, test_host_guest_writes, test_hud_layout_row, test_invalidate_quick, test_libc_hooks, test_linux_preflight, test_linux_pressure, test_linux_update, test_log_open_failure, test_missing_nup, test_mods_size_table, test_pending_guest_import, test_pipeline_play_cache, test_prepare_cache_save, test_present_thread, test_profiling_mode, test_psychov35_fsr3_fg, test_readback_timeout_source, test_slow_frames, test_song_note_cue, test_switch_table, test_upscaler_rows, test_vcore_escape, test_walk_key. The 37th, test_vbytes, hit the runner's 600 s timeout on a busy PC (a timed run beside it); it reads none of the saves (a lift check); on main @ 9b82abf6 alone, with --timeout 1800, it passed in 1,399 s, past the default 600 s, so it times out on main too with the default (this branch changes no code it builds).
  • In the game, on this PC (headless): longshot_ctl.py launch <game> --profile test --repo-save market-entrance --repo <a folder holding only this save's assets/saves/market-entrance> --headless into a new run folder copied only slot_00/ in; A on the title loaded it, and status read state play, map Market_Child, age child, no event, no load, 3 hearts, music BGM_Market, Link at [-1.74, 102.12, 64.0] facing -180 (into the Market); no scene played on the load. The committed files are byte for byte the ones loaded (a second scrub of the same slot gives the same bytes). The shot below is that load.

market-entrance, loaded with --repo-save: this PR @ 553b4a52

Not tested: Windows (the save is the game's own files, the same on both; launch --repo-save is covered by #543's tests).

Must be tested: None

Risks

What could break: Nothing in the game or the build: two save files and README text, read only by launch --repo-save and check-repo-saves. The saves' README table gains a row, so a sibling save PR (castle-gate) touches the same table and the second one merged takes a one-line rebase.

Turning it off: revert the commit.

Compatibility: Nothing changes for a player.

Checklist

Everyone

  • One topic.
  • The title is a Conventional Commit (type(scope): description) of the status line's type.
  • Why gives the reason plainly and links the issues that fit (the tracker searched).
  • Tested on my own PC where it runs (in the game, for a change the game runs), as Testing's Tested says; what
    can't be run by hand is in Testing's Not testable by hand, with why.
  • Measurements has a verdict, or one line saying why there is nothing to measure.
  • Nothing gets worse that I know of: any cost (speed, latency, memory, a setup that worked before) is measured and
    in Risks.
  • Testing names what was tested, what wasn't, and what must still be tested, where and how.
  • Rebased on current main.
  • python -I -B tests/run_all.py passes (any failure that also fails on main is named).
  • No game files, dumps, build output or generated files.
  • third_party/ untouched (a new folder with a PROVENANCE.md instead).
  • The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled.

When it applies

  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.
  • Screenshots of the same spot, captioned like the Measurements columns, for a change that shows on screen.
  • Docs updated.

🤖 Generated with Claude Code

**Type:** feat · **Risk:** low · **Tested on:** CachyOS Linux, RTX 3090 (NVIDIA 620.70.06) · **Needs:** none ## TL;DR The Test Farm can only start a run from Link's house in Kokiri Forest, so a test that walks into Castle Town's market had to teleport and walk there first. This adds a repository save, `market-entrance`: child Link in the Market's gate, the town ahead of him, by day, its first-visit scene already seen. It was made from a fresh profile and scrubbed like `kokiri-start`; it passes `check-repo-saves` and loads there in the game. ## Breakdown - **Who it helps:** the Test Farm and anyone testing a change in Castle Town's market, which is one of the game's busiest places. - **What was wrong:** the only repository save is in Kokiri Forest; a teleport reaches the Market, but a run that starts with a load there is the same on every PC and skips the first-visit scene. - **What changes:** a new save, `assets/saves/market-entrance`, and its row in the saves' README. - **What does not change:** the game, the scripts, `kokiri-start`; nothing for a player. - **How it was checked:** `check-repo-saves` passes; launched with `--repo-save market-entrance`, it loads in the Market's gate (state play, child, no event). - **What is still open:** nothing. - **Risk:** low: two save files and README text, read only by `launch --repo-save` and `check-repo-saves`. ## Why The Test Farm's planner asked for a town-entry save for tests of the render targets in the Market (#1255, #1256): `kokiri-start` is the only repository save, so a run had to teleport there itself, and the first walk in plays the Market's first-visit scene. With this save a run starts in the Market's gate, by day, the scene already seen, in the same place on every PC. The tracker has no issue for it (searched: "repo save", "assets/saves"). ## What changed A new repository save, `market-entrance`, made fresh from a new profile and scrubbed with `longshot_saves.py scrub`, and its row in the saves' README. - `assets/saves/market-entrance/slot_00/`: `progress.sav` (the game's save, unchanged) and `caption.sav` (its picture mid-grey, its date 2026-01-01 00:00). No `account.sav` or `option.sav`. - `assets/saves/market-entrance/README.md`: where the save stands (its first line) and how it was made. - `assets/saves/README.md`: the save's row in the table. How it was made, on this PC (CachyOS Linux, RTX 3090, NVIDIA 620.70.06; Longshot v3.0.0-rc.7-111-gda91e7702, main da91e770): a new profile with only `kokiri-start` in it (`longshot_ctl.py launch <game> --profile test --repo-save kokiri-start --headless`, an empty options file so every setting is at its default and the Autosave row On), Continue on the title, then through the control bridge `teleport Market_Child` (Link arrives in the Market's gate facing the town; that arrival writes no save: the save trace logged the requests and no slot changed), the left stick held back for 240 frames (out through the gate into Hyrule Field: an autosave, slot_01, the `castle-gate` save of its own PR), then back for 120 frames (through the gate into the Market again). The Market's first-visit scene played by itself, and the game's own autosave written after it (slot_02: scene Market, map Market/Market_Child, 3 hearts, by day) is this save; loading it puts Link where the teleport does, the scene already seen. Nothing else was done in the game; the story is a new game's, as in `kokiri-start`. Scrubbed on this PC, so the original picture never left it: ``` caption.sav: the save date (month at 280): 11 -> 1 caption.sav: the save date (day at 312): 19 -> 1 caption.sav: the save date (hour at 328): 2 -> 0 caption.sav: the save date (minute at 304): 16 -> 0 caption.sav: the save date (second at 288): 39 -> 0 caption.sav: the picture at 1396 (JPEG 543x306, 38241 bytes) replaced by a mid-grey JPEG 543x306 of 3345 bytes caption.sav: 39637 -> 4741 bytes; the rest of its header kept as it was (scene Market, map Market/Market_Child) progress.sav: copied unchanged ``` The save holds no game data: `check-repo-saves` passes, and `progress.sav`'s readable strings are only the game's own names (`Market_Child`, `Goto_HyruleCastle`, `Item_Wallet_Small` and the like), with no paths or names from the PC. ## Measurements Nothing to measure: this adds a save and a README row, no code; it is checked by `check-repo-saves`, the saves' tests and a load in the game (Testing). ### Flags and settings None ## Testing **Tested:** - `python -I scripts/longshot_saves.py check-repo-saves`: passes. - `tests/test_longshot_saves.py` (77 tests, OK, 1 skipped) and `tests/test_longshot_ctl.py` (51 tests, OK), on Linux. - `python -I -B tests/run_all.py` on this PC (Linux, a jj workspace of this branch on `main` @ 9b82abf6, before the rebase onto 99a7e249, which touched nothing here): 692 passed, 37 failed, 97 skipped. 36 of the failures also fail on `main` @ 9b82abf6 in a clean workspace (`run_all.py --exact` on them: 36 of 36 fail; they need tools or files that workspace lacks): `test_bug_report`, `test_bug_report_defaults`, `test_capture_read_once`, `test_crash_groups`, `test_crt_options`, `test_damage_debug_menu`, `test_dev_packs`, `test_display_startup_failure`, `test_dlss_feature_size`, `test_failed_compile_keeps_objects`, `test_guest_imported`, `test_hardware_report`, `test_high_fps_fixes`, `test_host_guest_writes`, `test_hud_layout_row`, `test_invalidate_quick`, `test_libc_hooks`, `test_linux_preflight`, `test_linux_pressure`, `test_linux_update`, `test_log_open_failure`, `test_missing_nup`, `test_mods_size_table`, `test_pending_guest_import`, `test_pipeline_play_cache`, `test_prepare_cache_save`, `test_present_thread`, `test_profiling_mode`, `test_psychov35_fsr3_fg`, `test_readback_timeout_source`, `test_slow_frames`, `test_song_note_cue`, `test_switch_table`, `test_upscaler_rows`, `test_vcore_escape`, `test_walk_key`. The 37th, `test_vbytes`, hit the runner's 600 s timeout on a busy PC (a timed run beside it); it reads none of the saves (a lift check); on `main` @ 9b82abf6 alone, with `--timeout 1800`, it passed in 1,399 s, past the default 600 s, so it times out on `main` too with the default (this branch changes no code it builds). - In the game, on this PC (headless): `longshot_ctl.py launch <game> --profile test --repo-save market-entrance --repo <a folder holding only this save's assets/saves/market-entrance> --headless` into a new run folder copied only `slot_00/` in; A on the title loaded it, and `status` read state `play`, map `Market_Child`, age `child`, no event, no load, 3 hearts, music BGM_Market, Link at [-1.74, 102.12, 64.0] facing -180 (into the Market); no scene played on the load. The committed files are byte for byte the ones loaded (a second scrub of the same slot gives the same bytes). The shot below is that load. ![market-entrance, loaded with --repo-save: this PR @ 553b4a52](https://git.magicbean.dev/attachments/829829af-5eae-40e0-b0a6-b725417de0d8) **Not tested:** Windows (the save is the game's own files, the same on both; `launch --repo-save` is covered by #543's tests). **Must be tested:** None ## Risks **What could break:** Nothing in the game or the build: two save files and README text, read only by `launch --repo-save` and `check-repo-saves`. The saves' README table gains a row, so a sibling save PR (`castle-gate`) touches the same table and the second one merged takes a one-line rebase. **Turning it off:** revert the commit. **Compatibility:** Nothing changes for a player. ## Checklist ### Everyone - [x] One topic. - [x] The title is a Conventional Commit (`type(scope): description`) of the status line's type. - [x] Why gives the reason plainly and links the issues that fit (the tracker searched). - [x] Tested on my own PC where it runs (in the game, for a change the game runs), as Testing's Tested says; what can't be run by hand is in Testing's Not testable by hand, with why. - [x] Measurements has a verdict, or one line saying why there is nothing to measure. - [x] Nothing gets worse that I know of: any cost (speed, latency, memory, a setup that worked before) is measured and in Risks. - [x] Testing names what was tested, what wasn't, and what must still be tested, where and how. - [x] Rebased on current `main`. - [x] `python -I -B tests/run_all.py` passes (any failure that also fails on `main` is named). - [x] No game files, dumps, build output or generated files. - [x] `third_party/` untouched (a new folder with a `PROVENANCE.md` instead). - [x] The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled. ### When it applies - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. - [x] Screenshots of the same spot, captioned like the Measurements columns, for a change that shows on screen. - [x] Docs updated. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
feat(saves): market-entrance, child Link in Castle Town Market's gate, by day
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 14s
CI / Workflow lint (actionlint) (pull_request) Successful in 21s
Guide screenshots / Guide images and local UI source review (pull_request) Failing after 21s
Third-party parts / As their notes say (pull_request) Successful in 24s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 4s
553b4a52bb
A repository save for the Test Farm: made fresh from a profile holding only
kokiri-start (Continue, teleport Market_Child, a walk out through the gate
into Hyrule Field and back in), the game's own autosave after the Market's
first-visit scene, scrubbed with longshot_saves.py scrub. Its README says how
it was made.

Assisted-by: Claude Code
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Author
Owner

Tested in game: CachyOS Linux, RTX 3090, NVIDIA 620.70.06, headless: launch --profile test --repo-save market-entrance, A on the title: state play, map Market_Child, age child, no event, 3 hearts, Link in the Market's gate facing the town, by day, no scene on the load

Tested in game: CachyOS Linux, RTX 3090, NVIDIA 620.70.06, headless: launch --profile test --repo-save market-entrance, A on the title: state play, map Market_Child, age child, no event, 3 hearts, Link in the Market's gate facing the town, by day, no scene on the load
Author
Owner

The full run_all's one timeout, test_vbytes, ran on main alone with a longer timeout: it passes in 1,399 s, past the default 600 s, so it times out on main too. Testing updated; marking ready.

The full run_all's one timeout, test_vbytes, ran on main alone with a longer timeout: it passes in 1,399 s, past the default 600 s, so it times out on main too. Testing updated; marking ready.
HumanBean changed title from WIP: feat(saves): market-entrance, child Link in Castle Town Market's gate, by day to feat(saves): market-entrance, child Link in Castle Town Market's gate, by day 2026-11-19 07:59:43 +00:00
futondestroyer left a comment

Recommendation: Wait on current-main save validation and the red Guide screenshots check. The submitted save has documented fresh-profile provenance and the permitted one-slot layout.

Does it help: This gives the Test Farm a repeatable market-entrance starting point. Performance measurement does not apply.
Other systems: This adds a save and catalog entry. There is no runtime code change.
Switches: None.
Reviewed: 553b4a52 against 99a7e249. Checked the tree layout, catalog, and provenance README, and read the author's validation report. Nothing run.
Fixed in this review: Nothing.
Checked and fine: The patch adds only the README and regular progress/caption files in slot_00, plus the catalog entry. No profile settings or links. HumanBean documents an isolated profile seeded with the approved Kokiri save, the transition autosave after the first Market scene, and a child/day reload facing the Market with no replayed scene. The reported scrub and check-repo-saves passed, and the author smoke-loaded it on CachyOS with an RTX 3090.

Testing: The description check passed. The Guide screenshots check failed, and the source checks were skipped. Please establish whether the failure reproduces on the exact control and clear the merge gate. Current main changed the save launch tooling. On the rebased pair, refresh check-repo-saves and smoke-load market-entrance through the current documented repo-save path. Each save requires two distinct eligible maintainer approvals before merge.
Overlaps and merge order: #1722 adds its own save and touches the same catalog. Either can go first. The second rebase should keep both rows and the current documentation link.

**Recommendation: Wait on current-main save validation and the red Guide screenshots check.** The submitted save has documented fresh-profile provenance and the permitted one-slot layout. **Does it help:** This gives the Test Farm a repeatable market-entrance starting point. Performance measurement does not apply. **Other systems:** This adds a save and catalog entry. There is no runtime code change. **Switches:** None. **Reviewed:** 553b4a52 against 99a7e249. Checked the tree layout, catalog, and provenance README, and read the author's validation report. Nothing run. **Fixed in this review:** Nothing. **Checked and fine:** The patch adds only the README and regular progress/caption files in slot_00, plus the catalog entry. No profile settings or links. HumanBean documents an isolated profile seeded with the approved Kokiri save, the transition autosave after the first Market scene, and a child/day reload facing the Market with no replayed scene. The reported scrub and check-repo-saves passed, and the author smoke-loaded it on CachyOS with an RTX 3090. **Testing:** The description check passed. The [Guide screenshots check](https://git.magicbean.dev/MagicBeanDev/Longshot/actions/runs/39610/jobs/0) failed, and the source checks were skipped. Please establish whether the failure reproduces on the exact control and clear the merge gate. Current main changed the save launch tooling. On the rebased pair, refresh check-repo-saves and smoke-load market-entrance through the current documented repo-save path. Each save requires two distinct eligible maintainer approvals before merge. **Overlaps and merge order:** https://git.magicbean.dev/MagicBeanDev/Longshot/pulls/1722 adds its own save and touches the same catalog. Either can go first. The second rebase should keep both rows and the current documentation link.
HumanBean deleted branch feat/save-market-entrance 2026-11-21 05:34:30 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
2 participants
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
MagicBeanDev/Longshot!1723
No description provided.