@farmslot/adapter-rn 0.31.0 → 0.31.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +8 -0
- package/package.json +4 -4
package/CHANGELOG.md
CHANGED
|
@@ -6,6 +6,14 @@ All notable changes to `@farmslot/adapter-rn` (published as `@farmslot/expo-reci
|
|
|
6
6
|
|
|
7
7
|
- Active-development baseline; add user-facing changes here before release or package publication.
|
|
8
8
|
|
|
9
|
+
## 0.31.2 - 2026-11-04
|
|
10
|
+
|
|
11
|
+
- Align published dependencies with the native delegation and recording-tail release cohort.
|
|
12
|
+
|
|
13
|
+
## 0.31.1 - 2026-11-03
|
|
14
|
+
|
|
15
|
+
- Align published dependency pins with the coordinated protocol, recorded-evidence and duplicate-timestamp timeline fixes.
|
|
16
|
+
|
|
9
17
|
## 0.31.0 - 2026-11-02
|
|
10
18
|
|
|
11
19
|
- `app.performance_capture` measures native UI frames on the slot's Android device from `adb shell dumpsys gfxinfo <app> framestats`. `phase: "start"` and `phase: "end"` bound a named window. The end writes `performance/<id>-summary.json`: the frame summary; the lowest FPS over any 500 ms of continuous rendering (recipe-runner's `lowestWindowedFps`, frames up to 100 ms apart), so a slow cadence or steady slow frames fail; and the latency of every `ui.press` and `ui.key_press`, overall and per target. A press in the end node's `shown_markers` runs from the frame that handled the touch past its screen's `[farmslot-perf] shown <screen>` logcat line (streamed with `adb logcat -v monotonic`, the frames' clock), which the app logs after React commits the screen, e.g. from the second `requestAnimationFrame` after focus: a press ends at the end of the last frame that started by two vsyncs after the screen's post-commit marker, or at the marker when that is later. It can over-read by up to ~2 light frames, and could still miss a mount frame that starts more than two vsyncs after the marker (not seen in 24 Pixel 6 presses); the report says so. No line before the next press leaves it untimed. Any other press, and every press in a window whose app logged no such line (noted), runs to the first frame that stands out as its mount within 600 ms, or past that up to 900 ms, after a pause too, a frame with a whole frame budget of UI-thread work; or to the first frame when none does. Devices before Android 12 leave presses untimed and say why. Given a `budget`, the window gets a verdict: failures name the metric, the limit, the measured value, the target and, given a `baseline`, its value. A coverage gap (failed polls, frames gfxinfo overwrote), or frames with no refresh budget, make it `inconclusive` whatever else failed, never a pass. `targets` are reported, never failed on. `<id>-frames.json` keeps the frames the latencies were read from. `phase: "report"` writes one report with the device, the app build and the checkout's HEAD, and fails when a window broke its budget or is inconclusive. A project enables it by declaring the action in its manifest with `"execution_capabilities": []`.
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@farmslot/adapter-rn",
|
|
3
|
-
"version": "0.31.
|
|
3
|
+
"version": "0.31.2",
|
|
4
4
|
"type": "module",
|
|
5
5
|
"main": "dist/index.js",
|
|
6
6
|
"types": "dist/index.d.ts",
|
|
@@ -51,9 +51,9 @@
|
|
|
51
51
|
"prepublishOnly": "node ../../scripts/quality/check-farmslot-package-readiness.mjs --publish --packages @farmslot/adapter-rn"
|
|
52
52
|
},
|
|
53
53
|
"dependencies": {
|
|
54
|
-
"@farmslot/agent-runtime": "0.
|
|
55
|
-
"@farmslot/protocol": "0.
|
|
56
|
-
"@farmslot/recipe-runner": "0.46.
|
|
54
|
+
"@farmslot/agent-runtime": "0.40.0",
|
|
55
|
+
"@farmslot/protocol": "0.76.0",
|
|
56
|
+
"@farmslot/recipe-runner": "0.46.6",
|
|
57
57
|
"pngjs": "^7.0.0",
|
|
58
58
|
"ws": "^8.18.0"
|
|
59
59
|
},
|