IDL: Remove UNLIKELY() from constructor

IDL: Remove UNLIKELY() from constructor

To remove a FIXME comment. I ran Dromaeo and didn't see any regression.
Here is the result.

w/ CL
1st: 1646.81runs/s (Total)
http://dromaeo.com/?id=224905

2nd: 1621.15runs/s (Total)
http://dromaeo.com/?id=224906

w/o CL
1st: 1662.16runs/s (Total)
http://dromaeo.com/?id=224907

2nd: 1644.60runs/s (Total)
http://dromaeo.com/?id=224908

Also, remove an obsolated FIXME comment.

BUG=345503
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=178934

[bashi, 2014-07-25 10:38:50]

Haraken-san, PTAL?

[haraken, 2014-07-25 10:42:37]

LGTM

[bashi, 2014-07-25 10:44:16]

The CQ bit was checked by bashi@chromium.org

[commit-bot: I haz the power, 2014-07-25 10:44:46]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/bashi@chromium.org/419673003/1

[commit-bot: I haz the power, 2014-07-25 13:53:11]

Change committed as 178934

[Nils Barth (inactive), 2014-07-25 14:12:50]

On 2014/07/25 13:53:11, I haz the power (commit-bot) wrote:
> Change committed as 178934

Are you sure this is ok?
Could you check again, collecting more data?

w/ CL:
1st: 1646.81runs/s (Total)
2nd: 1621.15runs/s (Total)
(avg: 1633.98)

w/o CL:
1st: 1662.16runs/s (Total)
2nd: 1644.60runs/s (Total)
(avg: 1653.38)

That looks about 1% slower!
Admittedly, there's a 1% test-to-test variation, but could you collect more data
to be sure?

Visually it's:
[...]....
...[...].
(i.e., some overlap, but looks different)

This is something we'd expect to only have a marginal impact, but it looks like
it might be measurable.

[Inactive, 2014-07-25 14:29:44]

For the record, I have also found those UNLIKELY() to have more impact on
ARM/Android. I am assuming you guys are testing Desktop?

[bashi, 2014-07-25 14:38:11]

On 2014/07/25 14:29:44, Chris Dumez wrote:
> For the record, I have also found those UNLIKELY() to have more impact on
> ARM/Android. I am assuming you guys are testing Desktop?

Yes. I ran the test on Linux desktop. I don't have strong opinion here(just
wanted to remove the FIXME comment), and if you experienced regressions on
mobile, we should revert the change. I'll prepare the revert.

[haraken, 2014-07-25 15:23:57]

On 2014/07/25 14:38:11, bashi1 wrote:
> On 2014/07/25 14:29:44, Chris Dumez wrote:
> > For the record, I have also found those UNLIKELY() to have more impact on
> > ARM/Android. I am assuming you guys are testing Desktop?
> 
> Yes. I ran the test on Linux desktop. I don't have strong opinion here(just
> wanted to remove the FIXME comment), and if you experienced regressions on
> mobile, we should revert the change. I'll prepare the revert.

Thanks for catching this! It makes sense to revert the change.