Rewrite std::string("") to std::string(), Linux edition.

Rewrite std::string("") to std::string(), Linux edition.

This patch was generated by running the empty_string clang tool
across the Chromium Linux compilation database. Implicitly or
explicitly constructing std::string() with a "" argument is
inefficient as the caller needs to emit extra instructions to
pass an argument, and the constructor needlessly copies a byte
into internal storage. Rewriting these instances to simply call
the default constructor appears to save ~14-18 kilobytes on an
optimized release build.

BUG=none

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=193020

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=193040

[dcheng, 2013-04-02 23:51:16]

Mind giving an overall opinion on this CL? It implements the change discussed in
the previous mail thread about std::string("") vs std::string(). This is the
result from running the tool in
https://src.chromium.org/viewvc/chrome?view=rev&revision=191936 and applying two
fixups:
1. clang-format using the instructions at
http://clang.llvm.org/docs/ClangFormat.html to avoid line length errors.
2. Manual fixups for most vexing parse errors and instances where the tool
rewrote L"" to std::wstring().

I think the questions of most interest are:
1) Should we be making the changes in the test files? "" is a bit more concise
to read and it doesn't really matter there. But if we allow that, there's a
strange inconsistency between the test files and Chrome proper.
2) Should I revert the third_party/ changes from this patch and work on
upstreaming them (and then pulling them back into Chrome)? I'm not quite sure
how these particular third_party dependencies are managed, since they don't
appear to be pulled in via DEPS.

[darin (slow to review), 2013-04-03 05:29:12]

On Tue, Apr 2, 2013 at 4:51 PM, <dcheng@chromium.org> wrote:

> Reviewers: darin,
>
> Message:
> Mind giving an overall opinion on this CL? It implements the change
> discussed in
> the previous mail thread about std::string("") vs std::string(). This is
> the
> result from running the tool in
>
https://src.chromium.org/**viewvc/chrome?view=rev&**revision=191936<https://s...
applying two
> fixups:
> 1. clang-format using the instructions at
>
http://clang.llvm.org/docs/**ClangFormat.html<http://clang.llvm.org/docs/Clan...
avoid line length errors.
> 2. Manual fixups for most vexing parse errors and instances where the tool
> rewrote L"" to std::wstring().
>
> I think the questions of most interest are:
> 1) Should we be making the changes in the test files? "" is a bit more
> concise
> to read and it doesn't really matter there. But if we allow that, there's a
> strange inconsistency between the test files and Chrome proper.
>

Hmm... it's a very good point.  The brevity in test files is useful to help
minimize line wrapping.  I suppose we could make the presubmit check permit
"" in test files.



> 2) Should I revert the third_party/ changes from this patch and work on
> upstreaming them (and then pulling them back into Chrome)? I'm not quite
> sure
> how these particular third_party dependencies are managed, since they don't
> appear to be pulled in via DEPS.
>

Yes, third_party should be handled separately and ignored by presubmit.
 Maybe third_party already is ignored by presubmit?



>
> Description:
> Rewrite most instances of std::string("") to std::string().
>
> Note that constructing std::string in an initializer with ""
> is not covered by this patch, since the Clang tool is not yet
> intelligent enough to handle elegant deletion of initializers.
>
>
OK

I guess that also means that a clang based presubmit would permit those too?

Thanks for working on this!
-Darin



> BUG=none
>
>
> Please review this at
https://codereview.chromium.**org/13145003/<https://codereview.chromium.org/1...
>
> SVN Base:
svn://svn.chromium.org/chrome/**trunk/src<http://svn.chromium.org/chrome/trunk/src>
>
> Affected files:
>   M base/debug/trace_event_**unittest.cc
>   M base/file_util_unittest.cc
>   M base/files/file_path.cc
>   M base/i18n/char_iterator_**unittest.cc
>   M base/json/json_value_**serializer_unittest.cc
>   M base/metrics/field_trial.cc
>   M base/metrics/field_trial_**unittest.cc
>   M base/metrics/statistics_**recorder.cc
>   M base/metrics/stats_counters.cc
>   M base/metrics/stats_table.cc
>   M base/nix/mime_util_xdg.cc
>   M base/pickle_unittest.cc
>   M base/prefs/pref_change_**registrar_unittest.cc
>   M base/prefs/pref_service_**unittest.cc
>   M base/process_util_linux.cc
>   M base/process_util_unittest.cc
>   M base/string_util_unittest.cc
>   M base/strings/string_split.cc
>   M base/strings/string_split_**unittest.cc
>   M base/sys_info_posix.cc
>   M base/test/trace_event_**analyzer.cc
>   M base/test/trace_event_**analyzer_unittest.cc
>   M base/values_unittest.cc
>   M base/vlog_unittest.cc
>   M chrome/browser/about_flags_**unittest.cc
>   M chrome/browser/accessibility/**accessibility_events.cc
>   M chrome/browser/autofill/**autofill_browsertest.cc
>   M chrome/browser/automation/**automation_misc_browsertest.cc
>   M chrome/browser/automation/**testing_automation_provider.cc
>   M chrome/browser/bookmarks/**bookmark_utils_unittest.cc
>   M chrome/browser/browsing_data/**browsing_data_cookie_helper_**
> unittest.cc
>   M chrome/browser/browsing_data/**browsing_data_database_helper_**
> browsertest.cc
>   M chrome/browser/browsing_data/**browsing_data_database_helper_**
> unittest.cc
>   M chrome/browser/browsing_data/**cookies_tree_model_unittest.cc
>   M chrome/browser/certificate_**manager_model.cc
>   M chrome/browser/content_**settings/content_settings_**browsertest.cc
>   M chrome/browser/content_**settings/content_settings_**
> internal_extension_provider.cc
>   M chrome/browser/content_**settings/content_settings_**
> origin_identifier_value_map_**unittest.cc
>   M chrome/browser/content_**settings/content_settings_**
> policy_provider.cc
>   M chrome/browser/content_**settings/content_settings_**
> policy_provider_unittest.cc
>   M chrome/browser/content_**settings/content_settings_**pref_provider.cc
>   M chrome/browser/content_**settings/content_settings_**
> pref_provider_unittest.cc
>   M chrome/browser/content_**settings/content_settings_**provider.h
>   M chrome/browser/content_**settings/content_settings_**
> provider_unittest.cc
>   M chrome/browser/content_**settings/content_settings_**utils.cc
>   M chrome/browser/content_**settings/content_settings_**utils_unittest.cc
>   M chrome/browser/content_**settings/cookie_settings.cc
>   M chrome/browser/content_**settings/host_content_**settings_map.cc
>   M chrome/browser/content_**settings/host_content_**
> settings_map_unittest.cc
>   M chrome/browser/content_**settings/mock_settings_**observer.cc
>   M chrome/browser/devtools/**browser_list_tabcontents_**provider.cc
>   M chrome/browser/devtools/**devtools_file_helper.cc
>   M chrome/browser/devtools/**devtools_sanity_browsertest.cc
>   M chrome/browser/download/**download_browsertest.cc
>   M chrome/browser/download/**download_file_picker.cc
>   M chrome/browser/download/**download_prefs.cc
>   M chrome/browser/extensions/**activity_log.cc
>   M chrome/browser/extensions/**activity_log_unittest.cc
>   M chrome/browser/extensions/api/**alarms/alarms_api_unittest.cc
>   M chrome/browser/extensions/api/**bookmarks/bookmarks_api.cc
>   M chrome/browser/extensions/api/**content_settings/content_**
> settings_apitest.cc
>   M chrome/browser/extensions/api/**content_settings/content_**
> settings_helpers.cc
>   M chrome/browser/extensions/api/**content_settings/content_**
> settings_store_unittest.cc
>   M chrome/browser/extensions/api/**cookies/cookies_api.cc
>   M chrome/browser/extensions/api/**cookies/cookies_helpers.cc
>   M chrome/browser/extensions/api/**cookies/cookies_unittest.cc
>   M chrome/browser/extensions/api/**declarative/initializing_**
> rules_registry.cc
>   M chrome/browser/extensions/api/**declarative/rules_registry_**
> with_cache_unittest.cc
>   M chrome/browser/extensions/api/**declarative_content/content_**
> condition.cc
>   M chrome/browser/extensions/api/**declarative_content/content_**
> rules_registry.cc
>   M chrome/browser/extensions/api/**declarative_content/**
> declarative_content_apitest.cc
>   M chrome/browser/extensions/api/**declarative_webrequest/**
> webrequest_rules_registry.cc
>   M chrome/browser/extensions/api/**developer_private/entry_**picker.cc
>   M chrome/browser/extensions/api/**dial/dial_device_data_**unittest.cc
>   M chrome/browser/extensions/api/**dial/dial_service_unittest.cc
>   M chrome/browser/extensions/api/**downloads/downloads_api.cc
>   M chrome/browser/extensions/api/**downloads/downloads_api_**unittest.cc
>   M chrome/browser/extensions/api/**extension_action/browser_**
> action_apitest.cc
>   M chrome/browser/extensions/api/**extension_action/page_action_**
> apitest.cc
>   M chrome/browser/extensions/api/**extension_action/script_badge_**
> apitest.cc
>   M chrome/browser/extensions/api/**file_system/file_system_api.cc
>   M chrome/browser/extensions/api/**file_system/file_system_api_**
> unittest.cc
>   M chrome/browser/extensions/api/**identity/identity_apitest.cc
>   M chrome/browser/extensions/api/**management/management_api_**
> browsertest.cc
>   M chrome/browser/extensions/api/**messaging/message_service.cc
>   M chrome/browser/extensions/api/**proxy/proxy_api.cc
>   M chrome/browser/extensions/api/**proxy/proxy_api_helpers_**unittest.cc
>   M chrome/browser/extensions/api/**socket/udp_socket.cc
>   M chrome/browser/extensions/api/**storage/settings_sync_**unittest.cc
>   M chrome/browser/extensions/api/**web_request/web_request_api.cc
>   M chrome/browser/extensions/api/**web_request/web_request_api_**
> helpers.cc
>   M chrome/browser/extensions/api/**web_request/web_request_api_**
> unittest.cc
>   M chrome/browser/extensions/api/**web_socket_proxy_private/web_**
> socket_proxy_private_api.cc
>   M chrome/browser/extensions/api/**webstore_private/webstore_**
> private_api.cc
>   M chrome/browser/extensions/api/**webstore_private/webstore_**
> private_apitest.cc
>   M chrome/browser/extensions/**bundle_installer.cc
>   M chrome/browser/extensions/**component_loader.cc
>   M chrome/browser/extensions/**component_loader_unittest.cc
>   M chrome/browser/extensions/**event_router.cc
>   M chrome/browser/extensions/**event_router_forwarder.cc
>   M chrome/browser/extensions/**extension_action_unittest.cc
>   M chrome/browser/extensions/**extension_apitest.cc
>   M chrome/browser/extensions/**extension_browsertest.h
>   M chrome/browser/extensions/**extension_browsertest.cc
>   M chrome/browser/extensions/**extension_function_dispatcher.**cc
>   M chrome/browser/extensions/**extension_pref_value_map_**unittest.cc
>   M chrome/browser/extensions/**extension_prefs.cc
>   M chrome/browser/extensions/**extension_process_manager.cc
>   M chrome/browser/extensions/**extension_service_unittest.cc
>   M chrome/browser/extensions/**extension_test_message_**listener.cc
>   M chrome/browser/extensions/**extensions_quota_service.cc
>   M chrome/browser/extensions/**external_policy_loader_**unittest.cc
>   M chrome/browser/extensions/**isolated_app_browsertest.cc
>   M chrome/browser/extensions/**menu_manager.cc
>   M chrome/browser/extensions/**pending_extension_info.cc
>   M chrome/browser/extensions/**platform_app_browsertest_util.**cc
>   M chrome/browser/extensions/**platform_app_launcher.cc
>   M chrome/browser/extensions/**script_badge_controller_**unittest.cc
>   M chrome/browser/extensions/**script_executor.cc
>   M chrome/browser/extensions/**subscribe_page_action_**browsertest.cc
>   M chrome/browser/extensions/tab_**helper.cc
>   M chrome/browser/extensions/**updater/extension_downloader.**cc
>   M chrome/browser/extensions/**updater/extension_updater.cc
>   M chrome/browser/extensions/**updater/extension_updater_**unittest.cc
>   M chrome/browser/extensions/**webstore_installer.cc
>   M chrome/browser/extensions/**webstore_standalone_installer.**cc
>   M chrome/browser/feedback/**feedback_util.cc
>   M chrome/browser/file_select_**helper.cc
>   M chrome/browser/google/google_**util.cc
>   M chrome/browser/google/google_**util_unittest.cc
>   M chrome/browser/google_apis/**base_operations.cc
>   M chrome/browser/google_apis/**base_operations_unittest.cc
>   M chrome/browser/google_apis/**drive_api_operations.cc
>   M chrome/browser/google_apis/**drive_api_operations_unittest.**cc
>   M chrome/browser/google_apis/**drive_api_url_generator_**unittest.cc
>   M chrome/browser/google_apis/**drive_uploader_unittest.cc
>   M chrome/browser/google_apis/**fake_drive_service_unittest.cc
>   M chrome/browser/google_apis/**gdata_wapi_operations_**unittest.cc
>   M chrome/browser/google_apis/**gdata_wapi_parser_unittest.cc
>   M chrome/browser/google_apis/**gdata_wapi_url_generator_**unittest.cc
>   M chrome/browser/google_apis/**operation_util_unittest.cc
>   M chrome/browser/google_apis/**test_server/http_server_**unittest.cc
>   M chrome/browser/history/**history_querying_unittest.cc
>   M chrome/browser/history/**history_tab_helper.cc
>   M chrome/browser/history/**history_unittest.cc
>   M chrome/browser/history/**thumbnail_database_unittest.cc
>   M chrome/browser/importer/**firefox_importer_utils.cc
>   M chrome/browser/internal_auth_**unittest.cc
>   M chrome/browser/io_thread.cc
>   M chrome/browser/language_usage_**metrics_unittest.cc
>   M chrome/browser/logging_chrome_**browsertest.cc
>   M chrome/browser/managed_mode/**managed_user_service.cc
>   M chrome/browser/media/media_**stream_devices_controller.cc
>   M chrome/browser/media_**galleries/media_galleries_**
> dialog_controller.cc
>   M chrome/browser/media_**galleries/media_galleries_**preferences.cc
>   M chrome/browser/metrics/**metrics_log.cc
>   M chrome/browser/metrics/**metrics_service.cc
>   M chrome/browser/nacl_host/**pnacl_file_host_unittest.cc
>   M chrome/browser/net/chrome_**fraudulent_certificate_**
> reporter_unittest.cc
>   M chrome/browser/net/chrome_**network_delegate_unittest.cc
>   M chrome/browser/net/dns_probe_**job_unittest.cc
>   M chrome/browser/net/http_**pipelining_compatibility_**
> client_unittest.cc
>   M chrome/browser/net/pref_proxy_**config_tracker_impl_unittest.**cc
>   M chrome/browser/net/url_info.cc
>   M chrome/browser/page_cycler/**page_cycler.cc
>   M chrome/browser/password_**manager/native_backend_**kwallet_x.cc
>   M chrome/browser/password_**manager/native_backend_**
> kwallet_x_unittest.cc
>   M chrome/browser/plugins/plugin_**finder.cc
>   M chrome/browser/plugins/plugin_**metadata_unittest.cc
>   M chrome/browser/policy/cloud/**cloud_policy_refresh_**
> scheduler_unittest.cc
>   M chrome/browser/policy/cloud/**cloud_policy_service_unittest.**cc
>   M chrome/browser/policy/cloud/**device_management_service_**unittest.cc
>   M chrome/browser/policy/cloud/**mock_cloud_policy_client.cc
>   M chrome/browser/policy/cloud/**mock_device_management_**service.cc
>   M chrome/browser/policy/**configuration_policy_pref_**store_unittest.cc
>   M chrome/browser/policy/policy_**map_unittest.cc
>   M chrome/browser/policy/policy_**prefs_browsertest.cc
>   M chrome/browser/policy/url_**blacklist_manager_unittest.cc
>   M chrome/browser/popup_blocker_**browsertest.cc
>   M chrome/browser/predictors/**autocomplete_action_predictor.**cc
>   M chrome/browser/predictors/**resource_prefetch_predictor.cc
>   M chrome/browser/predictors/**resource_prefetch_predictor_**
> tables_unittest.cc
>   M chrome/browser/predictors/**resource_prefetch_predictor_**unittest.cc
>   M chrome/browser/predictors/**resource_prefetcher_unittest.**cc
>   M chrome/browser/prefs/command_**line_pref_store_unittest.cc
>   M chrome/browser/prefs/proxy_**config_dictionary.cc
>   M chrome/browser/prefs/proxy_**policy_unittest.cc
>   M chrome/browser/prefs/session_**startup_pref.cc
>   M chrome/browser/prerender/**prerender_contents.cc
>   M chrome/browser/prerender/**prerender_histograms.cc
>   M chrome/browser/printing/cloud_**print/cloud_print_setup_flow.**cc
>   M chrome/browser/printing/cloud_**print/cloud_print_setup_**source.cc
>   M chrome/browser/printing/print_**dialog_cloud_interative_**uitest.cc
>   M chrome/browser/printing/print_**dialog_cloud_unittest.cc
>   M chrome/browser/profiles/gaia_**info_update_service_factory.cc
>   M chrome/browser/profiles/gaia_**info_update_service_unittest.**cc
>   M chrome/browser/profiles/**profile_downloader_unittest.cc
>   M chrome/browser/profiles/**profile_impl.cc
>   M chrome/browser/profiles/**profile_manager.cc
>   M chrome/browser/referrer_**policy_browsertest.cc
>   [[ 472 additional files ]]
>
>
>

[Peter Kasting, 2013-04-03 05:37:43]

On 2013/04/03 05:29:12, darin wrote:
> On Tue, Apr 2, 2013 at 4:51 PM, <mailto:dcheng@chromium.org> wrote:
> > 1) Should we be making the changes in the test files? "" is a bit more
> > concise
> > to read and it doesn't really matter there. But if we allow that, there's a
> > strange inconsistency between the test files and Chrome proper.
> >
> 
> Hmm... it's a very good point.  The brevity in test files is useful to help
> minimize line wrapping.  I suppose we could make the presubmit check permit
> "" in test files.

This argument applies equally to test and non-test files.  I don't think it's
compelling.  Test files should not use ""/L"" (except where necessary, e.g. when
returning a char*) any more than non-test files should.

[darin (slow to review), 2013-04-03 05:59:12]

On Tue, Apr 2, 2013 at 10:37 PM, <pkasting@chromium.org> wrote:

> On 2013/04/03 05:29:12, darin wrote:
>
>  On Tue, Apr 2, 2013 at 4:51 PM, <mailto:dcheng@chromium.org> wrote:
>> > 1) Should we be making the changes in the test files? "" is a bit more
>> > concise
>> > to read and it doesn't really matter there. But if we allow that,
>> there's a
>> > strange inconsistency between the test files and Chrome proper.
>> >
>>
>
>  Hmm... it's a very good point.  The brevity in test files is useful to
>> help
>> minimize line wrapping.  I suppose we could make the presubmit check
>> permit
>> "" in test files.
>>
>
> This argument applies equally to test and non-test files.  I don't think
> it's
> compelling.  Test files should not use ""/L"" (except where necessary,
> e.g. when
> returning a char*) any more than non-test files should.
>
>
Yes, but the downside to using a string literal is that it bloats the code.
 The bloat is inconsequential for tests, so it seems like the brevity
benefit then wins.  In contrast, bloat is something we are trying to avoid
for Chromium, even though the bloat we are talking about here is modest.
 So, for code that goes into Chromium, the loss of brevity seems acceptable.

-Darin



>
https://codereview.chromium.**org/13145003/<https://codereview.chromium.org/1...
>

[Peter Kasting, 2013-04-03 20:24:57]

On 2013/04/03 05:59:12, darin wrote:
> Yes, but the downside to using a string literal is that it bloats the code.

And takes longer to run, and misleads the reader into thinking that something
takes a char* that really takes a string.  The last is the most important.

>  The bloat is inconsequential for tests, so it seems like the brevity
> benefit then wins.

Far more important than any of these is consistency.  It is 1000x more important
that we not have inconsistent patterns between normal and testing code than that
we save a half dozen characters occasionally.  This is also why we require
things like out-of-line member function definitions even in testing code,
because we are trying to keep all rules consistent between the two.

[darin (slow to review), 2013-04-03 20:38:40]

I buy the consistency argument.  OK!

[Mark Mentovai, 2013-04-03 21:08:38]

I agree: no exception for tests.

[dcheng, 2013-04-03 21:12:24]

On 2013/04/03 21:08:38, Mark Mentovai wrote:
> I agree: no exception for tests.

Alright, sounds good to me. With that in mind, I'll finalize this initial fixup
patch and revert the third_party patches (I'll try to get them upstreamed, if
anyone has any tips on this, please let me know).
I'll be porting the "" detection over to the Clang plugin shortly after this,
and this patch will (hopefully) help provide some data towards that change.

[Nico, 2013-04-08 02:47:00]

(CL description needs a "why?" part)

[darin (slow to review), 2013-04-09 05:27:57]

I will trust in the great powers of your script.  LGTM

I note a couple examples of silly "std::string x = std::string()" that could be
improved.  Maybe a separate script run across the code to eliminate those could
be implemented.  OK to do that as a separate CL.

https://codereview.chromium.org/13145003/diff/38001/base/json/json_value_seri...
File base/json/json_value_serializer_unittest.cc (right):

https://codereview.chromium.org/13145003/diff/38001/base/json/json_value_seri...
base/json/json_value_serializer_unittest.cc:199: std::string test_serialization
= std::string();
nit: gratuitous assignment.

https://codereview.chromium.org/13145003/diff/38001/base/nix/mime_util_xdg.cc
File base/nix/mime_util_xdg.cc (right):

https://codereview.chromium.org/13145003/diff/38001/base/nix/mime_util_xdg.cc...
base/nix/mime_util_xdg.cc:349: std::string result = std::string();
nit: another one

[dcheng, 2013-04-09 05:37:04]

I've fixed up the std::string x = std::string() instances I found. I'll update
the tooling to be more clever about these in the future.

There's probably a few other things worth cleaning up; for example,
GURL(std::string()) shows up in a few places now.

[dcheng, 2013-04-09 05:45:36]

Committed patchset #5 manually as r193020.

[dcheng, 2013-04-09 08:47:07]

Committed patchset #6 manually as r193040.

[tfarina, 2013-04-09 13:19:12]

Daniel, does this patch contains entries like:

std::string my_string = std::string()?

Or you manually fixed those entries to:

std::string my_string;

?

I've saw this when I ran the empty_string tool here.

[dcheng, 2013-04-09 15:59:30]

On 2013/04/09 13:19:12, tfarina wrote:
> Daniel, does this patch contains entries like:
> 
> std::string my_string = std::string()?
> 
> Or you manually fixed those entries to:
> 
> std::string my_string;
> 
> ?
> 
> I've saw this when I ran the empty_string tool here.

There should be no instances of = std::string() in this patch.

[scottmg, 2013-04-15 17:03:44]

Was this CL tested for size changes on Windows?

http://build.chromium.org/f/chromium/perf/xp-release/sizes/report.html?histor...
shows a regression of about 180K in the range [193035..193064] for chrome.dll.

The only other plausible candidate appears to be a NaCl roll, but that claimed
to reduce binary size too, and I'm not sure it'd affect chrome.dll size in any
case.

(I'm starting a test now, but the opt build takes a long time, so I thought I'd
confirm if this has already been done)

[dcheng, 2013-04-15 17:12:54]

On 2013/04/15 17:03:44, scottmg wrote:
> Was this CL tested for size changes on Windows?
> 
>
http://build.chromium.org/f/chromium/perf/xp-release/sizes/report.html?histor...
> shows a regression of about 180K in the range [193035..193064] for chrome.dll.
> 
> The only other plausible candidate appears to be a NaCl roll, but that claimed
> to reduce binary size too, and I'm not sure it'd affect chrome.dll size in any
> case.
> 
> (I'm starting a test now, but the opt build takes a long time, so I thought
I'd
> confirm if this has already been done)

I'm fairly confident this is the NaCl roll. If you look at the output from the
other sizes bots, sizes went down after this patch (finally) landed and only
increased after the NaCl roll hit the tree.
https://groups.google.com/a/chromium.org/forum/?fromgroups=#!search/native$20...
also comments on how the uncompressed size of the new nacl validator is much
larger, and the nacl roll in question picked up the new validator.

[scottmg, 2013-04-15 17:56:02]

On 2013/04/15 17:12:54, dcheng wrote:
> On 2013/04/15 17:03:44, scottmg wrote:
> > Was this CL tested for size changes on Windows?
> > 
> >
>
http://build.chromium.org/f/chromium/perf/xp-release/sizes/report.html?histor...
> > shows a regression of about 180K in the range [193035..193064] for
chrome.dll.
> > 
> > The only other plausible candidate appears to be a NaCl roll, but that
claimed
> > to reduce binary size too, and I'm not sure it'd affect chrome.dll size in
any
> > case.
> > 
> > (I'm starting a test now, but the opt build takes a long time, so I thought
> I'd
> > confirm if this has already been done)
> 
> I'm fairly confident this is the NaCl roll. If you look at the output from the
> other sizes bots, sizes went down after this patch (finally) landed and only
> increased after the NaCl roll hit the tree.
>
https://groups.google.com/a/chromium.org/forum/?fromgroups=#%21search/native%...
> also comments on how the uncompressed size of the new nacl validator is much
> larger, and the nacl roll in question picked up the new validator.

Thanks, I misinterpreted the NaCl PSA. Logged
https://code.google.com/p/chromium/issues/detail?id=231445.