Bug 54019 – http/tests/websocket/tests/frame-length-skip.html is flaky

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Mario Sanchez Prada]

This layout test is randomly failing every now and then in the GTK 32 bit release bot at least.

Looks like it's flaky. Perhaps the following 2 secs timeout is not enough sometimes:

   [...]

   ws.onclose = function () {
       debug("WebSocket is closed");
       finish();
   };
   var timeoutID = setTimeout("finish()", 2000);

   [...]

...but that's just my wild guessing, of course :-)


The diff output for build #10514 in Gtk 32 bit release bot is as follows:

--- /var/lib/buildbot/build/gtk-linux-32-release/build/layout-test-results/http/tests/websocket/tests/frame-length-skip-expected.txt	2011-02-08 10:00:02.000000000 -0800
+++ /var/lib/buildbot/build/gtk-linux-32-release/build/layout-test-results/http/tests/websocket/tests/frame-length-skip-actual.txt	2011-02-08 10:00:02.000000000 -0800
@@ -2,12 +2,8 @@
 
 On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
 
-WebSocket is open
-received:hello
-received:world
-WebSocket is closed
-2
-PASS areArraysEqual(received_messages, expected_messages) is true
+0
+FAIL areArraysEqual(received_messages, expected_messages) should be true. Was false.
 PASS successfullyParsed is true
 
 TEST COMPLETE


Not skipping, as it seems this is not an isolated problem with layout tests for websockets stuff.

Adding its author in CC, as well as some others webkitgtk gardeners.

[Eric Seidel (no email)]

Adam moved the websocket tests over to a real http server and most of the flakiness went away. If you see this again, please re-open.

[Philippe Normand]

Committed r85469: <http://trac.webkit.org/changeset/85469>

[Philippe Normand]

I meant r85470