Omnibox: Don't Call Classify() Repeatedly

Omnibox: Don't Call Classify() Repeatedly

Tested:
* Search for "qrty.com" repeatedly.  Verify that when you type "q", a search for "qrty.com" does not get inline autocompleted.  This is the same as the behavior before this changelist.  Without the Classify() code, this suppression of the inline autocompletion does not happen.
* Check speed.  Search for "testing 1", "testing 2", ... "testing 13", ...  i.e., create a lot of searches starting with "t".  Restart browser.  Open seven blank new tabs and two separate tabs of about:histograms.  In each blank tab, type "t"; don't hit return.  Reload one of the histograms tabs.  Compare the Omnibox.ProviderTime.Search histogram in the  non-reloaded histograms tab with the reloaded histograms tab.
  - in current chrome (before this patch), I see a difference in the laggy section of the histogram.  In my testing, I see 7 new entries in the 129-204ms bucket.  Ugh.
  - in chrome built with this patch, I see a difference in the 13-21ms bucket; no different in higher buckets.

BUG=393956, 262263
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=285456

[Mark P, 2014-07-23 23:52:40]

Peter,

What do you think of this approach?  I think it should work well enough at
speeding things up.  I'm not too worried about the quality impact.  I can
justify myself nonchalance to you if you want (in person is easiest; I started
typing reasons here but decided you might not care), or you can ask me to
measure it if you are concerned.  At the moment I'm guessing this change has a
tiny quality impact.

Note: I haven't yet tested it to confirm that fixes the lagginess isssues.

--mark

https://codereview.chromium.org/412063003/diff/1/chrome/browser/autocomplete/...
File chrome/browser/autocomplete/search_provider.cc (right):

https://codereview.chromium.org/412063003/diff/1/chrome/browser/autocomplete/...
chrome/browser/autocomplete/search_provider.cc:946: // The 1200 relevance score
threshold in the test below is the lowest
I added stuff here to the comment.

https://codereview.chromium.org/412063003/diff/1/chrome/browser/autocomplete/...
chrome/browser/autocomplete/search_provider.cc:957: //  * When the user has
typed the whole string before as a query, then it's
I rewrote (re-justified) this second bullet point in the comment.

[Peter Kasting, 2014-07-23 23:55:53]

LGTM

[Mark P, 2014-07-24 19:08:14]

I have tested this and it fixes the speed issue.

Now submitting.

Please correct me if your "LGTM" was only a "the approach looks good" rather
than "the code itself looks good".

--mark

[Mark P, 2014-07-24 19:08:25]

The CQ bit was checked by mpearson@chromium.org

[commit-bot: I haz the power, 2014-07-24 19:09:12]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mpearson@chromium.org/412063003/1

[Mark P, 2014-07-24 19:33:59]

FYI, I updated the changelist description to include the testing I did.

--mark

[commit-bot: I haz the power, 2014-07-24 22:46:33]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  linux_chromium_chromeos_rel on tryserver.chromium
(http://build.chromium.org/p/tryserver.chromium/builders/linux_chromium_chrome...)
  linux_chromium_rel_swarming on tryserver.chromium
(http://build.chromium.org/p/tryserver.chromium/builders/linux_chromium_rel_sw...)
  mac_chromium_rel on tryserver.chromium
(http://build.chromium.org/p/tryserver.chromium/builders/mac_chromium_rel/buil...)

[commit-bot: I haz the power, 2014-07-25 02:29:50]

Change committed as 285456