Make chrome.omnibox.setDefaultSuggestion work when called inside

Make chrome.omnibox.setDefaultSuggestion work when called inside
onInputChanged.

BUG=64538
TEST=Use the chrome_search sample extension and make sure the first suggestion updates in sync with what you type.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=68411

[Peter Kasting, 2010-12-06 20:05:53]

LGTM

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
File chrome/browser/autocomplete/keyword_provider.cc (right):

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:447: break;
Nit: Using "return" instead of "break" makes life a little easier on the reader
(I don't have to scan through the rest of the code looking for post-switch
functionality).  (3 places)

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:454:
!ExtractKeywordFromInput(input, &keyword, &remaining_input))
Is ExtractKeywordFromInput() supposed to guaranteed-succeed here, like in
EXTENSION_OMNIBOX_SUGGESTIONS_READY below?

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:471: // TODO(mpcomplete):
consider clamping the number of suggestions to
Nit: Doesn't this TODO belong on the next block?  Also, why not go ahead and
implement this?

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:481: suggestions.suggestions[i];
Nit: You could probably avoid this temp by using an iterator in the loop.  (Then
it would be "i->" instead of "suggestion.".)

[Matt Perry, 2010-12-07 00:10:18]

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
File chrome/browser/autocomplete/keyword_provider.cc (right):

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:447: break;
On 2010/12/06 20:05:53, Peter Kasting wrote:
> Nit: Using "return" instead of "break" makes life a little easier on the
reader
> (I don't have to scan through the rest of the code looking for post-switch
> functionality).  (3 places)

Done.

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:454:
!ExtractKeywordFromInput(input, &keyword, &remaining_input))
On 2010/12/06 20:05:53, Peter Kasting wrote:
> Is ExtractKeywordFromInput() supposed to guaranteed-succeed here, like in
> EXTENSION_OMNIBOX_SUGGESTIONS_READY below?

I *think* it should as long as the first 2 clauses of the conditional are false,
but I wasn't entirely sure about all the edge cases.

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:471: // TODO(mpcomplete):
consider clamping the number of suggestions to
On 2010/12/06 20:05:53, Peter Kasting wrote:
> Nit: Doesn't this TODO belong on the next block?  Also, why not go ahead and
> implement this?

I've moved the TODO. I think I'm going to wait until
http://code.google.com/p/chromium/issues/detail?id=56705 is fixed before doing
anything here.

http://codereview.chromium.org/5607003/diff/2001/chrome/browser/autocomplete/...
chrome/browser/autocomplete/keyword_provider.cc:481: suggestions.suggestions[i];
On 2010/12/06 20:05:53, Peter Kasting wrote:
> Nit: You could probably avoid this temp by using an iterator in the loop. 
(Then
> it would be "i->" instead of "suggestion.".)

I also need the index for the relevance calculation below.