initialize v8 early in unittests

initialize v8 early in unittests

BUG=
R=jkummerow@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=24256

[dcarney, 2014-09-26 14:11:42]

Patchset #1 (id:1) has been deleted

[dcarney, 2014-09-26 14:11:50]

Patchset #1 (id:20001) has been deleted

[Jakob Kummerow, 2014-09-26 14:14:35]

jkummerow@chromium.org changed reviewers:
+ jkummerow@chromium.org

[Jakob Kummerow, 2014-09-26 14:14:35]

LGTM with nit.

https://codereview.chromium.org/608773002/diff/40001/src/test/run-all-unittes...
File src/test/run-all-unittests.cc (right):

https://codereview.chromium.org/608773002/diff/40001/src/test/run-all-unittes...
src/test/run-all-unittests.cc:42: // This forces some thread local key
initialization that may be need to print
nit: s/need/needed/

[dcarney, 2014-09-26 14:17:12]

Committed patchset #1 (id:40001) manually as 24256 (tree was closed).

[Benedikt Meurer, 2014-09-29 04:24:30]

bmeurer@chromium.org changed reviewers:
+ bmeurer@chromium.org

[Benedikt Meurer, 2014-09-29 04:24:31]

Seriously this has to be reverted. This is exactly the opposite of what we're
trying to accomplish, i.e. do initialization of V8 only when requested and do it
during test case/fixture setup. Jochen is almost there with his V8::Initialize
changes.

[Benedikt Meurer, 2014-09-29 04:32:25]

On 2014/09/29 at 04:24:31, Benedikt Meurer wrote:
> Seriously this has to be reverted. This is exactly the opposite of what we're
trying to accomplish, i.e. do initialization of V8 only when requested and do it
during test case/fixture setup. Jochen is almost there with his V8::Initialize
changes.

And of course, printing test names should never depend on the state of V8, so
the right fix seems to be quite easy.