Cleanup: Remove some unneeded string allocations.

Cleanup: Remove some unneeded string allocations.

The string conversion functions called all take StringPieces.

Committed: https://crrev.com/75f873577a040fb0ed5c2e895a0650c62771662f
Cr-Commit-Position: refs/heads/master@{#306679}

[Lei Zhang, 2014-12-03 20:42:44]

thestig@chromium.org changed reviewers:
+ brettw@chromium.org

[brettw, 2014-12-03 21:12:21]

lgtm

https://codereview.chromium.org/776903002/diff/20001/chrome/browser/safe_brow...
File
chrome/browser/safe_browsing/incident_reporting/environment_data_collection_win.cc
(right):

https://codereview.chromium.org/776903002/diff/20001/chrome/browser/safe_brow...
chrome/browser/safe_browsing/incident_reporting/environment_data_collection_win.cc:127:
module_state->set_name(base::WideToUTF8(modules_to_verify[i]));
Note that this is the same since there's no wide string piece. But in this case
the implicit constructor works and looks cleaner, so this type of change seems
fine.

https://codereview.chromium.org/776903002/diff/20001/components/user_manager/...
File components/user_manager/empty_user_info.cc (left):

https://codereview.chromium.org/776903002/diff/20001/components/user_manager/...
components/user_manager/empty_user_info.cc:20: return
base::UTF8ToUTF16(std::string());
This old code is pretty bad.

[Lei Zhang, 2014-12-03 21:37:52]

The CQ bit was checked by thestig@chromium.org

[commit-bot: I haz the power, 2014-12-03 21:39:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/776903002/20001

[commit-bot: I haz the power, 2014-12-03 21:42:35]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2014-12-03 21:43:31]

Patchset 2 (id:??) landed as
https://crrev.com/75f873577a040fb0ed5c2e895a0650c62771662f
Cr-Commit-Position: refs/heads/master@{#306679}

[brucedawson, 2014-12-17 00:37:16]

I'm looking at logging.cc where an explicit cast to std::wstring was removed by
this code, but this does not actually change the generated code because
base::WideToUTF8 does not actually have a StringPiece overload. All the change
did (for this source file) was to change from an explicit conversion to an
implicit conversion. Presumably this applies to the other nine places where a
call to base::WideToUTF8 was changed.

If this was an important part of the optimization then adding a StringPiece
overload would be worthwhile.

[Lei Zhang, 2014-12-17 01:01:16]

On 2014/12/17 00:37:16, brucedawson wrote:
> I'm looking at logging.cc where an explicit cast to std::wstring was removed
by
> this code, but this does not actually change the generated code because
> base::WideToUTF8 does not actually have a StringPiece overload. All the change
> did (for this source file) was to change from an explicit conversion to an
> implicit conversion. Presumably this applies to the other nine places where a
> call to base::WideToUTF8 was changed.
> 
> If this was an important part of the optimization then adding a StringPiece
> overload would be worthwhile.

Yes, I realize there isn't a performance improvement there due to
base::WideToUTF8() not taking a StringPiece. I didn't see that until after the
fact, so the commit message should have said mostly instead of all. It's also to
simplify code, and to get rid of a bad pattern to prevent proliferation via
copy+pasting. In an extreme case, one bad pattern ended up in 10+ files because
people kept using the file and its copies as templates for other files in the
same directory.