testbed: stop destroying the player's save

The documented way to run the testbed began with

    rm -rf ~/AppData/Roaming/LOVE/Alchemist

which is the player's actual playthrough. It was written down as the
official procedure in four places, so every run of the test suite ate a
real save -- including twice in the session that produced the previous
commit. Reported by Talon.

The testbed does genuinely need an empty save directory: it drives a new
game from the main menu through the prologue, which has already happened
in any real save. So tests/run-testbed.sh MOVES the save aside (never
deletes it), runs, discards the test run's directory and moves the real
one back. Restore is on an EXIT trap plus INT/TERM, so it also happens on
a failed run, a crash and Ctrl-C. If the restore itself fails it says so
loudly and leaves the backup on disk rather than pressing on.

This cannot live inside the mod: love.load calls config.initialize,
keybindings.initialize and user.initialize long before main.lua reaches
require("accessibility").initialize(), so by the time any of our Lua runs
the save directory has already been read. The swap has to happen around
the process.

The rm -rf is removed from all four docs that carried it, each replaced
with the script and the reason not to clear that directory by hand --
being written down as procedure is what made it recur.

accessibility/.gitattributes forces LF on *.sh, or a Windows checkout
hands bash a \r and the script dies. It sits inside accessibility/ on
purpose: a top-level one would show up in the game-file footprint check,
which still reads `main.lua | 1 +`.

Verified with a marker file in a populated save directory: 38 checks, 0
failures, marker intact, no leftover backup directories.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Accessibility Mod
2026-08-02 02:59:34 +02:00
co-authored by Claude Opus 5
parent 0df3842bf1
commit c841bcc68d
5 changed files with 180 additions and 17 deletions
+58 -11
View File
@@ -2,12 +2,12 @@
**Read this first on any handoff.** Living document: update it whenever state changes, before ending a turn.
Last updated: 2026-08-02 · Currently: **Phases 0–5 COMPLETE. Handed off for
playtesting.** Remaining work is polish, mod settings and distribution.
Last updated: 2026-08-02 · Currently: **Phases 0–5 COMPLETE. In playtesting** —
fixing what Talon hits. Remaining work is polish, mod settings and distribution.
**Game is playable start-to-finish by ear** from main menu through the intro
conversation into gameplay, plus journal, character, actions and inventory —
verified by the testbed, **34/34 checks**.
verified by the testbed, **38/38 checks**.
**Not set up yet, or unsure how the two trees relate?** →
[00-dev-setup.md](00-dev-setup.md).
@@ -15,13 +15,18 @@ verified by the testbed, **34/34 checks**.
### Run the testbed (do this after any change)
```
cd "D:/code/Sulphur Memories Alchemist 0.4.2 sanctified"
rm -rf ~/AppData/Roaming/LOVE/Alchemist # start from a clean save
"C:/Program Files/LOVE/lovec.exe" . --a11y-test
bash accessibility/tests/run-testbed.sh
```
⚠ **Use the script, not `lovec . --a11y-test`.** The testbed needs an empty save
directory, and this used to be a bare `rm -rf ~/AppData/Roaming/LOVE/Alchemist`
in these docs — which ate Talon's actual playthrough. The script *moves* the save
aside, runs, and moves it back (restore is on an EXIT trap, and it never deletes
the real one). It can't be done inside the mod: `love.load` reads config,
keybindings and user data long before `main.lua` reaches our entry point.
It drives the game by code via `love.event.push`, navigating by *reading* focus
rather than counting keypresses. Every bug listed below was caught by it.
⚠ **The check COUNT varies between runs (23–26); this is expected.** Several
⚠ **The check COUNT varies between runs; this is expected.** Several
blocks are conditional — the config-menu section only runs if `selectByText`
lands on "Options", and mouse-hover can move menu focus at startup (see trap 2).
**What matters is `0 failures`, not the total.**
@@ -565,6 +570,48 @@ invariant directly instead of trying to capture speech. **34/34 testbed checks**
`In view: a chest, 1 tile south-west. a torch on the wall, 3, nearest 2 tiles
south-west. …`.
### Done — inventory item action menu (bug fix, reported by Talon)
**Symptom:** enter/`z` on an inventory item makes a click and then nothing is
spoken. Equipping the tutorial torch was therefore impossible to confirm.
**Cause:** the click *is* the menu opening. `inventory.keypressed` →
`inventoryWindow.enableItemMenu()` → `modules/item/itemMenu.lua`, a **second,
nested selection** drawn as a small popup over the grid: the list of actions this
item currently allows, built by `modules/item/getActionsList.lua` from the
per-action `condition`s in `itemActions.lua` (Activate, Wield, Wear, Switch on,
Drink, Read, Open, Prepare for throwing…). Up/down move in it, enter activates,
**any other key cancels it**. None of this goes through `console.print` or the
cursor popup, so it was entirely silent. `TEsound.play(sound.select)` in
`enableItemMenu` was the click. **This is the game's only equipment UI** — there
is no separate equip screen — so the whole screen was unusable.
**Fix** — `accessibility/states/_itemgrid.lua`:
- `window.itemMenu` is checked **first** and the grid cell stays quiet under it;
the menu is modal over the grid, so announcing the cell would be wrong.
- The menu is a closure with no getters, so `lines` and `currentSelectedAction`
are read as upvalues of its own `draw()`.
- Leads with the item name only when the menu just opened. The "just opened"
memory is **the menu table itself**, not a flag, which keeps `read()`
idempotent (see the journal warning above — a flag would be consumed by the
poll and the real announcement would come out headless).
- **`isEquipped` is now part of the cell text and of the focus key.** After
"Switch on" the menu closes with the cursor unmoved, so without it the re-read
would be suppressed as unchanged and the action would still feel silent. Now:
`Switch on` → `a torch, equipped, slot 3`.
**Plus `accessibility/patches/itemmenu.lua`** (6th patch), wrapping the
`views/inventoryWindow` factory: when `getActionsList` returns nil — an item with
no applicable actions, e.g. a plain reagent — the game does *literally nothing*,
not even the click, which by ear is indistinguishable from a dropped keypress.
Says `<item>, no actions`. Success is not announced here; the descriptor owns it.
⚠ **Still not exercised against a real item.** The prologue inventory is empty,
so the testbed instead builds an `itemMenu` directly from a fake usable item and
asserts the two upvalue names still resolve — the version-bump canary for this
descriptor. **38/38 checks.** The live path (an actual torch, actually equipped)
is Talon's to confirm.
### Known gaps (next up)
- **Phase 5**: player status key (HP, statuses, current action), popup
announcements (`popupStack.add`), damage alerts
@@ -629,7 +676,7 @@ utterance — speech is verifiable from the log, not only by ear. `!` after
## HANDOFF — read this if you are resuming
### State of play
Phases 0–5 complete. **23 state descriptors, 5 patches, 34/34 testbed checks**,
Phases 0–5 complete. **23 state descriptors, 6 patches, 38/38 testbed checks**,
history unit tests passing, **one line changed in the game** (`main.lua`).
Everything in the original brief is implemented: console speech with
@@ -639,8 +686,7 @@ exploration, wall tones, and keyboard+speech access to every screen.
### First thing to do after any change
```
lua accessibility/tests/test_history.lua # pure Lua, instant
rm -rf ~/AppData/Roaming/LOVE/Alchemist # clean save
"C:/Program Files/LOVE/lovec.exe" . --a11y-test # ~4 min, 34 checks
bash accessibility/tests/run-testbed.sh # ~4 min, 38 checks
git diff --stat vanilla-0.4.2 -- . ':!accessibility' ':!.gitignore' ':!CLAUDE.md' # must stay: main.lua | 1 +
```
@@ -672,8 +718,9 @@ These are blocked on content the prologue does not contain, **not** known broken
- **Tracker categories** Items / Alchemy / Exits / Creatures / Terrain. Only
Interactables is populated in the tutorial house. Diagnostic: the testbed
prints `tracker contents:` with per-category kind names.
- **Inventory item names** — inventory is empty in the prologue, so only the
"empty slot" path has run.
- **Inventory item names, the `equipped` suffix and the item ACTION MENU** —
inventory is empty in the prologue, so only the "empty slot" path has run live.
The action-menu upvalue names are covered by a synthetic testbed check.
- **The quantity slider** (`patches/slider.lua`). Confirmed *applied* by the
self-test but never exercised: `ingame_wait` is unreachable because
`tutorial_bed` has an empty `tryInteractFunction` by design, and