test: skip flaky parts of broadcastchannel test on Windows · nodejs/node@22028aa · GitHub

nodejs / node Public

• Notifications You must be signed in to change notification settings
• Fork 36.6k
• Star 120k

• Code
• Issues 638
• Pull requests 748
• Actions
• Projects
• Security and quality
• Insights

Additional navigation options

Commit 22028aa

Browse filesBrowse the repository at this point in the historyBrowse files

Trott

authored and

danielleadams

committed

test: skip flaky parts of broadcastchannel test on Windows

The feature was added recently and is experimental. This will need to be fixed before it can be released as stable (in my opinion at least). But for now, this gets us to a green CI without skipping the parts of the test that are working on Windows. PR-URL: #36386 Reviewed-By: Beth Griggs <bgriggs@redhat.com> Reviewed-By: Richard Lau <rlau@redhat.com>

1 parent 4b4caad commit 22028aa
Copy full SHA for 22028aa

1 file changed

+5-11Lines changed: 5 additions & 11 deletions

File tree

Expand file treeCollapse file tree

Open diff view settings

Filter options

• test/parallel
  • test-worker-broadcastchannel-wpt.js

Expand file treeCollapse file tree

Open diff view settings

Collapse file

‎test/parallel/test-worker-broadcastchannel-wpt.js‎

Copy file name to clipboardExpand all lines: test/parallel/test-worker-broadcastchannel-wpt.js

+5-11Lines changed: 5 additions & 11 deletions

Original file line number	Diff line number	Diff line change
@@ -27,7 +27,7 @@ const {
27	27	// dispatching algorithm under the covers. What's not
28	28	// immediately clear is whether the ordering is spec
29	29	// mandated. In this test, c1 should receive events
30		- // first, then c2, then c3. In the Node.js non-Windows dispatching
	30	+ // first, then c2, then c3. In the Node.js dispatching
31	31	// algorithm this means the ordering is:
32	32	// from c3 (c1 from c3)
33	33	// done (c1 from c2)
@@ -36,7 +36,7 @@ const {
36	36	// from c1 (c3 from c1)
37	37	// done (c3 from c2)
38	38	//
39		- // Whereas in Windows and in the browser-ordering (as illustrated in the
	39	+ // Whereas in the browser-ordering (as illustrated in the
40	40	// Web Platform Tests) it would be:
41	41	// from c1 (c2 from c1)
42	42	// from c1 (c3 from c1)
@@ -56,14 +56,9 @@ const {
56	56	doneCount++;
57	57	if (doneCount === 2) {
58	58	assert.strictEqual(events.length, 6);
59		- if (common.isWindows) {
60		- assert.strictEqual(events[0].data, 'from c1');
61		- assert.strictEqual(events[2].data, 'from c1');
62		- assert.strictEqual(events[3].data, 'from c3');
63		- assert.strictEqual(events[3].data, 'from c3');
64		- assert.strictEqual(events[5].data, 'done');
65		- assert.strictEqual(events[5].data, 'done');
66		- } else {
	59	+ // TODO: Don't skip Windows once ordering is fixed per comment above.
	60	+ // Right now, the ordering for Windows is unreliable.
	61	+ if (!common.isWindows) {
67	62	assert.strictEqual(events[0].data, 'from c3');
68	63	assert.strictEqual(events[1].data, 'done');
69	64	assert.strictEqual(events[2].data, 'from c1');
@@ -132,7 +127,6 @@ const {
132	127	{
133	128	// TODO: Fix failure on Windows CI. Skipping for now.
134	129	if (!common.isWindows) {
135		-
136	130	// Closing a channel in onmessage prevents already queued tasks
137	131	// from firing onmessage events
138	132	const c1 = new BroadcastChannel('close-in-onmessage2').unref();

0 commit comments

Comments

0 (0)