Add function support for Sleep and MessageLoop with TimeDelta input.

Add function support for Sleep with TimeDelta input.

This is the first step of fixing issue 108171 (converting calls of Sleep() to
use TimeDelta instead of ints in milliseconds).  I checked
platform_thread_unittests.cc for any tests explicitly for Sleep but found none,
so I didn't add any for this interface.

There will be a bit more implementation juggling here once the Sleep(int ms)
interface is removed, but that's coming in a later CL.

BUG=108171
TEST=


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=116012

[Ted Vessenes, 2011-12-28 05:40:58]

I added interfaces for Sleep and MessageLoop with TimeDelta.  This is the first
step.  Next step will involve converting function calls to use TimeDelta inputs
instead of int64.  Last step is removing the int64 function interfaces.

Two caveats.  First, I didn't add unit tests for these changes.  As best I can
tell there's no tests for Sleep() (and I'm not even sure how you'd test it). 
The MessageLoop code has tests, but it wasn't clear to me at what stage they
should be added.  Should the current MessageLoop tests just get converted to use
TimeDelta instead, or should the tests get duplicated, and then the int64 tests
removed in the last CL?

Second, I had to change a bit of code in base/threading/thread_unittests.cc to
explicitly specify which function pointer to base::PlatformThread::Sleep to use
in a binding.  rsleevi mentioned that some kind of explicit templating for
Bind<void(int)> could be used instead of the static_cast<>, but I couldn't get
it working.  And since that will get removed in the final CL anyway, I didn't
sweat it.

[brettw, 2011-12-28 20:10:27]

http://codereview.chromium.org/8965072/diff/2001/base/message_loop.cc
File base/message_loop.cc (right):

http://codereview.chromium.org/8965072/diff/2001/base/message_loop.cc#newcode330
base/message_loop.cc:330: const tracked_objects::Location& from_here, const
base::Closure& task,
The closure should be on the next line.

http://codereview.chromium.org/8965072/diff/2001/base/threading/platform_thre...
File base/threading/platform_thread_posix.cc (right):

http://codereview.chromium.org/8965072/diff/2001/base/threading/platform_thre...
base/threading/platform_thread_posix.cc:172:
Sleep(duration.InMillisecondsRoundedUp());
It seems like this is backwards. Sleep(ms) should be implemented in terms of
this function. Both TimeDelta and the timespec used in the existing sleep code
have a lot more resolution than just ms. While it arguably doesn't matter for
this case, it seems weird to throw it away.

[Ted Vessenes, 2011-12-29 20:10:10]

I think it's reasonable to support microsecond level sleeping if that's what
people want to put in the TimeDelta, so I reversed the implementation.  Windows
seems to only support millisecond level sleeps, so that didn't change.

Incidentally, TimeDelta's resolution only goes up to microsecond while nanosleep
supports nanoseconds (obviously).  I thought about adding a
TimeDelta::InNanoseconds(), but decided to use InMicroseconds() * 1000 instead. 
Having an InNanoseconds() function might imply to users that the TimeDelta
object supports nanosecond level resolution, which it does not.  But I'm happy
to add that function if you think it's worthwhile to do so.

[brettw, 2011-12-29 23:16:45]

lgtm

http://codereview.chromium.org/8965072/diff/5001/base/threading/platform_thre...
File base/threading/platform_thread_posix.cc (right):

http://codereview.chromium.org/8965072/diff/5001/base/threading/platform_thre...
base/threading/platform_thread_posix.cc:169: sleep_time.tv_nsec =
duration.InMicroseconds() * 1000; // nanoseconds
Two spaces before end of line comments.

[commit-bot: I haz the power, 2011-12-29 23:47:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/tedvessenes@gmail.com/8965072/10001

[commit-bot: I haz the power, 2011-12-30 01:04:37]

Change committed as 116012