HACK on m21 branch 1180 to avoid flickering of buffered bar by lying.

HACK on m21 branch 1180 to avoid flickering of buffered bar by lying.
Instead of reporting multi-range buffered information, revert to reporting a
single [0,max) buffered range, where max is an unholy combination of
currentTime()+0.5s (to ensure the bar never disappears), duration() (to ensure
the bar doesn't overflow), and the real end of the buffered ranges from the
pipeline (to still retain some semblance of usefulness, when the bytes/time
ratio is stable enough).

BUG=133567

[Ami GONE FROM CHROMIUM, 2012-06-26 20:16:04]

scherkus: this is the branch-only hack I mentioned in IM.  WDYT?

[scherkus (not reviewing), 2012-06-27 17:52:18]

lgtm

https://chromiumcodereview.appspot.com/10680004/diff/1/webkit/media/webmediap...
File webkit/media/webmediaplayer_impl.cc (right):

https://chromiumcodereview.appspot.com/10680004/diff/1/webkit/media/webmediap...
webkit/media/webmediaplayer_impl.cc:485: WebKit::WebTimeRanges
web_ranges(static_cast<size_t>(1));
1u doesn't work!? :~(

[Ami GONE FROM CHROMIUM, 2012-06-27 18:15:19]

https://chromiumcodereview.appspot.com/10680004/diff/1/webkit/media/webmediap...
File webkit/media/webmediaplayer_impl.cc (right):

https://chromiumcodereview.appspot.com/10680004/diff/1/webkit/media/webmediap...
webkit/media/webmediaplayer_impl.cc:485: WebKit::WebTimeRanges
web_ranges(static_cast<size_t>(1));
On 2012/06/27 17:52:18, scherkus wrote:
> 1u doesn't work!? :~(

Nope.  The template foo in WebVector tries to treat "1" as a class type and
confuses the compiler.  Haven't dug into why.