Script streaming: add UMA for tracking the reason why we don't stream.

Script streaming: add UMA for tracking the reason why we don't stream.

This will make it possible to see whether it's possible to stream more scripts
than currently (and what needs to be done to do that).

Suspecting that the biggest reason for not streaming is that the resource is
already loaded (which is OK, no actions needed there).

BUG=

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=192362

[marja, 2015-03-23 12:23:53]

marja@chromium.org changed reviewers:
+ haraken@chromium.org

[marja, 2015-03-23 12:23:54]

haraken@, ptal. Some "streaming version 2" stuff. (Or rather, investigating
whether there's potential for enhancements... though, I suspect the main reason
for not streaming is that the resource is already loaded.)

[haraken, 2015-03-23 13:17:38]

LGTM

[marja, 2015-03-23 13:21:31]

The CQ bit was checked by marja@chromium.org

[commit-bot: I haz the power, 2015-03-23 13:21:49]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1031623002/1

[commit-bot: I haz the power, 2015-03-23 14:41:55]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=192362

[Ilya Sherman, 2015-03-23 22:14:43]

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

[Ilya Sherman, 2015-03-23 22:14:44]

https://codereview.chromium.org/1031623002/diff/1/Source/bindings/core/v8/Scr...
File Source/bindings/core/v8/ScriptStreamer.cpp (right):

https://codereview.chromium.org/1031623002/diff/1/Source/bindings/core/v8/Scr...
Source/bindings/core/v8/ScriptStreamer.cpp:496:
blink::Platform::current()->histogramEnumeration(kNotStreamingReasonHistogramName,
3, 8);
Could you please use enumerated constant names, rather than using opaque
constants?  One advantage is that if you ever need to update the list, it's a
lot less likely that the upper bound will be forgotten.

[marja, 2015-03-24 15:38:46]

Will do (in the near future). At some point I thought I'd just remove this UMA
after being done w/ the analysis, but actually, why not keep it in place for
monitoring that streaming keeps running...

According to the first numbers (only 3000 samples), ~50% are "script already
loaded" (as expected) and the other ~50% is "script too small", so, looks like
streaming works as expected all in all. Other reasons are negligible.