Chromium's faster, multi-threaded, python-based testing harness landed in svn.webkit.org about a week ago. On my 4 core work machine it runs all the tests in 5 minutes vs. 8 minutes with run-webkit-tests. On Dirk's 16 core machine it runs all the tests in < 2minutes. HUGE thanks to Dirk Pranke who did the work to get it into webkit.org. Huge thanks to Ojan and Pam and all the others who have worked on the testing infrastructure over the years. It still has a bunch of bugs and isn't anywhere near ready for general usage. It only works on Mac, likely only Leopard, you can try it with: run-chromium-webkit-tests --platform=mac-leopard (or mac-snow-leopard on leopard). If you're interested in following the progress you can CC yourself here: https://bugs.webkit.org/show_bug.cgi?id=34984 My goal is to get this python version up and working as well or better than the perl one and then slowly transition parts of WebKit's architecture over to use it. That's weeks, if not months away. -eric
On Feb 19, 2010, at 3:12 PM, Eric Seidel wrote:
Chromium's faster, multi-threaded, python-based testing harness landed in svn.webkit.org about a week ago.
On my 4 core work machine it runs all the tests in 5 minutes vs. 8 minutes with run-webkit-tests. On Dirk's 16 core machine it runs all the tests in < 2minutes.
HUGE thanks to Dirk Pranke who did the work to get it into webkit.org. Huge thanks to Ojan and Pam and all the others who have worked on the testing infrastructure over the years.
It still has a bunch of bugs and isn't anywhere near ready for general usage. It only works on Mac, likely only Leopard, you can try it with: run-chromium-webkit-tests --platform=mac-leopard (or mac-snow-leopard on leopard).
If you're interested in following the progress you can CC yourself here: https://bugs.webkit.org/show_bug.cgi?id=34984
My goal is to get this python version up and working as well or better than the perl one and then slowly transition parts of WebKit's architecture over to use it. That's weeks, if not months away.
I like the fasterness and think the new tool is a good basis for future work. Maybe it would be good to do a feature set comparison, so we can make sure our future layout test running tool does everything we need. Regards, Maciej
On Fri, Feb 19, 2010 at 6:27 PM, Maciej Stachowiak <mjs@apple.com> wrote:
I like the fasterness and think the new tool is a good basis for future work. Maybe it would be good to do a feature set comparison, so we can make sure our future layout test running tool does everything we need.
There are definitely some things that run-webkit-tests does that run-chromium-webkit-tests doesn't (and vice versa). I believe Eric is planning to do such a feature comparison as part of the alluded-to migration. -- Dirk
participants (3)
-
Dirk Pranke
-
Eric Seidel
-
Maciej Stachowiak