W3C home > Mailing lists > Public > public-web-perf@w3.org > March 2011

Recent testharness.js updates causing issues with webperf tests

From: Nic Jansma <Nic.Jansma@microsoft.com>
Date: Thu, 10 Mar 2011 02:08:25 +0000
To: "public-web-perf@w3.org" <public-web-perf@w3.org>
CC: "graham@opera.com" <graham@opera.com>
Message-ID: <F677C405AAD11B45963EEAE5202813BD19C5FCC5@TK5EX14MBXW652.wingroup.windeploy.ntdev.microsoft.com>
fyi:

There appears to be a different version of testharness.js on test.w3.org vs. w3c-test.org.  The test.w3.org version seems to be updated to the latest Mercurial version, while w3c-test.org is using a version from 25-Jan-2011.
http://test.w3.org/resources/testharness.js
http://w3c-test.org/resources/testharness.js

However, the /webperf/ folder on both sites appears to be current to our latest Mercurial changes.

It appears that a recent update to testharness.js (and thus published on test.w3.org) may be causing some issues with some of our tests.  It looks like asserts() that happen after the load handler for the page are not getting registered in the summary of Pass/Fail results.

For example, test.w3.org with the new changes shows "Found 0 tests" in IE and Chrome while the older version on w3c-test.org works properly and shows all of the Pass/Fail results:
http://test.w3.org/webperf/tests/approved/test_navigate_within_document.htm
http://w3c-test.org/webperf/tests/approved/test_navigate_within_document.htm

These approved tests used to work in both browsers fully.

I've reviewed some of the changes to testharness.js but haven't fully debugged what is causing the problem.

- Nic
Received on Thursday, 10 March 2011 02:09:01 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:04:30 UTC