M5: a chain of After delay variant changes plays to the end (#2352)

Change to swapped the instance's variant in place, so the next variant's After delay
kept the instance's node id and was already recorded as fired: the chain stopped
after one step. The swap now clears the fired keys inside the instance and records
when it happened; next_delay counts a timer from the later of the screen's arrival
and that swap.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Model-Effort: Opus 5.5 drove; Opus 5.5 briefed; Sonnet 5.5 wrote at medium
This commit is contained in:
Trevor BrownandClaude Sonnet 5.5 committed 2026-11-13 23:42:18 -05:00
1 parent 24613d3fa6
commit f451e60ceb
5 files changed
+113 -2

No files matched your search

+2 -1
View File
@@ -39,7 +39,7 @@ tool; opt-in), **silent** (the documentation doesn't say; the case pins the curr
| Key/gamepad on a controller: the trigger takes a controller button in place of a key, pressed anywhere on the screen | **added** | `Gamepad A` … `Gamepad Home` (the standard layout's 17 buttons, face buttons by position) as the trigger's key, picked from the Controller menu beside the key field; the player polls controllers only while the prototype has such a trigger and the window has the focus, and plays each press once (held buttons don't repeat); Windows reads Xbox-compatible controllers, the browser its Gamepad API's standard-mapped ones; `controller_buttons_fire_key_gamepad_triggers`, `controller_buttons_play_key_gamepad_triggers` |
| A controller button is not the keyboard key of the same name; the d-pad is not the arrow keys; a keyboard modifier held makes a button press another input | silent | pinned as stated |
| REST files: controller triggers (`device` other than `KEYBOARD`) | **added** | they were skipped; their code reads as the standard layout's button index (silent: the schema doesn't say what the codes count); two buttons at once or an unknown code is skipped with a warning; `controller_triggers_import_as_their_button` |
| After delay: timed from when the screen shows | pass | `triggers_fire_on_their_gesture` |
| After delay: timed from when the screen shows; a chain of variant changes on After delay plays to the end, each step's delay counted from when its variant appears (#2352) | pass | `triggers_fire_on_their_gesture`, `a_chain_of_variant_after_delays_plays_to_the_end` |
| After delay on an overlay: timed from when the overlay opens, again each time it opens | **fixed** | timed from when the screen under it arrived (a toast opened late closed at once) and fired only the first time; `an_overlays_after_delay_counts_from_when_it_opens` |
| Nested layers with the same trigger: the top-most layer's interaction wins; a layer without that trigger lets it through to the one under it | pass | `nested_hotspots_and_click_through`, `hotspots_priority_and_click_through` |
| A layer without interactions over a hotspot | silent | it doesn't block it: clicks reach the hotspot under it |
@@ -108,6 +108,7 @@ tool; opt-in), **silent** (the documentation doesn't say; the case pins the curr
| Date | Change | Summary |
|---|---|---|
| 2026-11-14 | patch | M5, #2352: a chain of variant changes on After delay plays to the end (A to B to C to D); before, the step a variant had fired stayed marked as fired when the instance changed variant, so the chain stopped after one step, and now each step's delay counts from when its variant appears (`crates/ui-egui/src/present/afterdelay_chain_tests.rs`). Numbers unchanged |
| 2026-11-09 | patch | M7, #2158: a connection dragged from a variant (or a layer in it) onto a sibling variant of its set did nothing, because the drop looked only for top-level frames and a variant's top-level frame is its set. It now creates Change to the sibling (On click, Instant, as measured in Figma's web app) and outlines the sibling while dragging; the set's background and the variant itself still add nothing, and a frame outside the set still gets Navigate to |
| 2026-11-06 | patch | M5, #2027: `proto.settings` took any string as the device (`"Pixel Nonexistent"`, `"iphone 16"`) and saved it in the file; the presenter then drew no device while its picker showed the name as if it were one, and the settings couldn't be read back over MCP (`proto.settings {}` answered `{ok: true}`). An unknown name (or a non-string) is now refused with every name the picker offers (`None`, `Custom size` and the devices), as an unknown colour is refused with the device's colours, and nothing changes; a known name is taken in any case and stored as the picker spells it. Every call answers the page's settings after it: device, background, rotation, the colour shown, the screen as presented, the custom size and responsive. Test: `proto_present_tests::unknown_devices_are_refused_and_settings_read_back` |
| 2026-10-31 | minor | Gap 15: the Controller menu names buttons as PlayStation and Switch controllers do; that item leaves the missing list. Estimate unchanged. Test: `present::tail_tests`. |