The open-source Arc replacement for macOS. Spaces, pinned tabs, split view, Little Windows and Arc's shortcuts on Chromium, with your Arc sidebar imported in one go.
Nightly scenarios run again: fixture server without a name lookup, release build, warm-up
Victor565e477c4d20
diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.ymlindex 24087db..4600656 100644--- a/.github/workflows/nightly.yml+++ b/.github/workflows/nightly.yml@@ -35,8 +35,10 @@ jobs:path: vendor/cefkey: cef-${{ runner.arch }}-${{ steps.cef.outputs.version }}-${{ hashFiles('scripts/fetch-cef.sh') }}- - name: Build (debug, ad-hoc signed)- run: scripts/build.sh debug+ # Release, as users run it: the scenario budgets (the recorder's 0.1 % of a core above all)+ # are a release build's. A debug build's recorder alone costs 0.16–0.22 %.+ - name: Build (release, ad-hoc signed)+ run: scripts/build.sh- name: Scenariosenv:@@ -49,6 +51,10 @@ jobs:echo "| Scenario | Result | Longest stall / over 50 ms | Memory | Problems |"echo "|---|---|---|---|---|"} >> "$GITHUB_STEP_SUMMARY"+ # A fresh runner's first Cra launch pages CEF in from disk on the main thread (1.1–2.7 s+ # at start-up, 3–6 Nov) whatever the scenario; one throwaway run first, so the start-up+ # budget measures Cra, not the runner's cold disk.+ scripts/scenario.sh solid-spaces > /dev/null 2>&1 || truescripts/scenario.sh all- name: Keep logsdiff --git a/docs/testing.md b/docs/testing.mdindex 56ec88b..3b36c2b 100644--- a/docs/testing.md+++ b/docs/testing.md@@ -64,6 +64,9 @@ no page content), so it checks geometry, ink and fills.its budget, helper processes back to baseline five seconds after tabs close, the performancerecorder under 0.1 % of a core (`selfCost.cpuPct` in its perf.json, `budget perfcpu=`) with itscount of 250 ms stalls within a few of HitchMonitor's, and a quit within 20 seconds.+ The budgets are for a release build (`scripts/build.sh`), which is what the nightly runs: a+ debug build's recorder costs about six times as much (0.16–0.22 % against 0.03–0.04 %, 6 Nov+ 2026), so on a debug build the recorder line fails and means nothing.Stall budgets are set from what the app does today, so they catch regressions rather thandescribe a goal. The count of stalls over 50 ms is steady from run to run and carries thediff --git a/scripts/scenario.sh b/scripts/scenario.shindex 6b6af88..0dc5dac 100755--- a/scripts/scenario.sh+++ b/scripts/scenario.sh@@ -58,7 +58,7 @@ fiNAME="${1:?usage: scripts/scenario.sh <name>|all}"FILE="Tests/Scenarios/$NAME.scenario"[ -f "$FILE" ] || { echo "No scenario $FILE"; exit 2; }-[ -x "$APP/Contents/MacOS/Cra" ] || { echo "No Cra at $APP (scripts/build.sh debug)"; exit 2; }+[ -x "$APP/Contents/MacOS/Cra" ] || { echo "No Cra at $APP (scripts/build.sh; budgets are a release build's)"; exit 2; }if grep -q '^requires screen' "$FILE" && [ "${CRA_SCENARIO_SCREEN:-}" != "1" ]; thenecho "$NAME: skipped (takes over the screen; CRA_SCENARIO_SCREEN=1 runs it)"; exit 0fi@@ -96,9 +96,19 @@ problem() { PROBLEMS+=("$1"); echo " ✗ $1"; }# The fixture pages over http on a free port: a real host, no network.FIXED_PORT=$(sed -n 's/^port \([0-9]*\).*/\1/p' "$FILE" | head -1)-python3 -I -u -m http.server --bind 127.0.0.1 --directory "$FIXTURES/Pages" "${FIXED_PORT:-0}" >"$OUT/server.log" 2>&1 &+# Not `python3 -m http.server`: it looks up 127.0.0.1's name before it says its port, and on a CI+# runner that lookup outlasted the wait for every scenario after the first (nightly, 3–6 Nov).+python3 -I -u -c '+import functools, http.server, socketserver, sys+class Server(http.server.ThreadingHTTPServer):+ def server_bind(self):+ socketserver.TCPServer.server_bind(self)+ self.server_name, self.server_port = "127.0.0.1", self.server_address[1]+s = Server(("127.0.0.1", int(sys.argv[2])), functools.partial(http.server.SimpleHTTPRequestHandler, directory=sys.argv[1]))+print(f"Serving HTTP on 127.0.0.1 port {s.server_address[1]}", flush=True)+s.serve_forever()' "$FIXTURES/Pages" "${FIXED_PORT:-0}" >"$OUT/server.log" 2>&1 &SERVER=$!-for _ in $(seq 1 50); do PORT=$(sed -n 's/.*port \([0-9]*\).*/\1/p' "$OUT/server.log" | head -1); [ -n "$PORT" ] && break; sleep 0.1; done+for _ in $(seq 1 300); do PORT=$(sed -n 's/.*port \([0-9]*\).*/\1/p' "$OUT/server.log" | head -1); [ -n "$PORT" ] && break; sleep 0.1; donePAGE="http://127.0.0.1:${PORT:?http server did not start}/page.html"# The same server under another host, which the run lists as a DRM service (CRA_WEBKIT_HOSTS):# its pages open in WebKit, the way Spotify and Netflix do (ProtectedMediaSites).