Revert "RegExp: Add support for the ES6-proposed sticky flag"

Revert "RegExp: Add support for the ES6-proposed sticky flag"

Causes a flaky failure on buildbots. Here is the (deterministic) repro step (thanks to Michael Stanton):

first go to flag-definitions.h and set this to false.

DEFINE_BOOL(enable_sse4_1, false,
            "enable use of SSE4.1 instructions if available")

Run the following and it should fail:

tools/run-tests.py --arch=ia32 --mode=release cctest/test-api/Regress2107

R=yangguo@chromium.org
BUG=

Committed: https://code.google.com/p/v8/source/detail?r=24045

[Yang, 2014-09-18 14:45:29]

On 2014/09/18 14:45:14, rossberg wrote:

lgtm.

[rossberg, 2014-09-18 14:53:16]

Committed patchset #1 manually as 24045 (tree was closed).

[Sven Panne, 2014-09-19 06:35:48]

On 2014/09/18 14:53:16, rossberg wrote:
> Committed patchset #1 manually as 24045 (tree was closed).

I think this was the wrong CL to revert, Regress2107 still fails:
http://chromegw.corp.google.com/i/client.v8/builders/V8%20Linux%20-%20deadcod...

My personal guess for the guilty commit would be
https://code.google.com/p/v8/source/detail?r=24037 or
https://code.google.com/p/v8/source/detail?r=24042...

[Erik Corry, 2014-09-19 06:52:21]

erik.corry@gmail.com changed reviewers:
+ erik.corry@gmail.com

[Erik Corry, 2014-09-19 06:52:21]

My bet is on https://codereview.chromium.org/576093002/ since it changes the
idle notification heuristics.

The test is congenitally flaky, since it makes calls to a heuristic guidance
method, idleNotification, and expects predictable results.  Change the
heuristics and it breaks, but that doesn't mean it wasn't broken to start with.