[webkit-dev] Removing Support for Python 2.5
abarth at webkit.org
Mon Nov 21 14:01:10 PST 2011
I'm happy to give the Chromium project as much time as it needs to
remove these dependencies on Python 2.5, but I'd like to set a
deadline so that the WebKit project doesn't need to support Python 2.5
indefinitely. Is three months sufficient?
(I'm happy to help make this happen on the Chromium side with my
Chromium hat on, but I probably won't be able to do it all myself.)
On Fri, Nov 18, 2011 at 10:09 AM, Tony Chang <tony at chromium.org> wrote:
> The bot step that runs ui_tests uses scripts/slave/runtest.py. You could
> probably set PATH in that script before launching ui_tests.
> On Thu, Nov 17, 2011 at 5:22 PM, Adam Barth <abarth at webkit.org> wrote:
>> (Sorry if we're spamming webkit-dev with this thread. We can move it
>> to a Chromium infrastructure mailing list if we're bugging anyone.)
>> On Thu, Nov 17, 2011 at 5:17 PM, Dirk Pranke <dpranke at chromium.org> wrote:
>> > On Thu, Nov 17, 2011 at 5:06 PM, Eric Seidel <eric at webkit.org> wrote:
>> >> On Thu, Nov 17, 2011 at 4:53 PM, Tony Chang <tony at chromium.org> wrote:
>> >>> new-run-webkit-httpd imports common/host.py which imports lots of
>> >>> stuff
>> >>> including common/net/buildbot.py, which will fail to import the json
>> >>> module.
>> >>> I would land smaller pieces so the rollback isn't as painful. :)
>> >> Adam appears to have listed the files:
>> >> http://codesearch.google.com/#search/&exact_package=chromium&q=new-run-webkit&type=cs
>> >> Can someone with chromium commit-bit, please move those 3 files to
>> >> call python 2.7 in whatever the proper way is, so we can move forward
>> >> here?
>> > 2.7? Aren't we just requiring 2.6? I don't think any of the Chromium
>> > bots can assume 2.7 (except for the ones Tony has upgraded), and none
>> > of my machines (except for maybe my Lion machine at home) even have
>> > 2.7 installed.
>> The pattern Tony used for new-run-webkit-tests was to detect whether
>> Python 2.7 was installed at a certain path before trying to use it.
>> That pattern seems like it would work in these cases too. The one
>> tricky one is ui_test because the launcher is in C++. Perhaps it
>> should call a wrapper script instead?
More information about the webkit-dev