All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=670734

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgx4KVrwkM


Bot(s) for this bug's original alert(s):

mac

Hi chenwilliam@, was https://codereview.chromium.org/2513423003 expected to increase binary size on any platform? (Size of GoogleChrome.app appears to have increased by about 1M.)

I think it's unlikely my changes caused this size regression because all of the C++ code changes were for the test_runner / content_shell which shouldn't affect the binary size of the main app. Perhaps something else changed? I can't look at the buildbot stdio page, but it might be helpful to see if something in stdio changed from the previous build: https://build.chromium.org/p/chromium.perf/builders/Mac%20Builder/builds/56839/steps/sizes/logs/stdio

+dnj: Any ideas if the logs for these builds are in logdog somewhere?
https://build.chromium.org/p/chromium.perf/builders/Mac%20Builder/builds/56838/steps/sizes/logs/stdio
https://build.chromium.org/p/chromium.perf/builders/Mac%20Builder/builds/56839/steps/sizes/logs/stdio

https://luci-logdog.appspot.com/v/?s=chrome/bb/chromium.perf/Mac_Builder/56838/+/recipes/steps/sizes/0/stdout
https://luci-logdog.appspot.com/v/?s=chrome/bb/chromium.perf/Mac_Builder/56839/+/recipes/steps/sizes/0/stdout

Feel free to generate your own URLs, the pattern should be obvious from here :)

Actually this might be easier:
https://luci-milo.appspot.com/logdog/build/chrome/bb/chromium.perf/Mac_Builder/56839/+/annotations

Make sure you log in, and the pattern here for other builds should also be obvious.

Thanks dnj@ - I reviewed the diff between the two builds and nothing jumps out.

+tapted, owner of Mac sizes. Any ideas why it could have jumped at chenwilliam's CL?

GoogleChrome.app is not just binary size. It's all the resources too.

resources.pak went from 17,020 kB to 18,224 kB between 57.0.2938.0 57.0.2939.0. That's too much.

https://codereview.chromium.org/2513423003 seems to bundle a lot of generated files from out/gen. Perhaps this is vulcanised/flattened js/html output that should be minified as well.



$ diff -W 180 --suppress-common-lines --side-by-side /tmp/2938.txt /tmp/2939.txt
92	./Google Chrome Framework.framework/Resources/app_mode_loader.app/Contents/Ma |	96	./Google Chrome Framework.framework/Resources/app_mode_loader.app/Contents/Ma
										      >	268	./Google Chrome Framework.framework/Resources/zh_TW.lproj/locale.pak
272	./Google Chrome Framework.framework/Resources/zh_TW.lproj/locale.pak	      <
										      >	296	./Google Chrome Framework.framework/Resources/da.lproj/locale.pak
300	./Google Chrome Framework.framework/Resources/da.lproj/locale.pak	      <
316	./Google Chrome Framework.framework/Resources/nl.lproj/locale.pak	      |	312	./Google Chrome Framework.framework/Resources/nl.lproj/locale.pak
										      >	328	./Google Chrome Framework.framework/Resources/es.lproj/locale.pak
332	./Google Chrome Framework.framework/Resources/es.lproj/locale.pak	      <
1052	./Google Chrome Framework.framework/Resources/snapshot_blob.bin		      |	1116	./Google Chrome Framework.framework/Resources/snapshot_blob.bin
17020	./Google Chrome Framework.framework/Resources/resources.pak		      |	18224	./Google Chrome Framework.framework/Resources/resources.pak
112976	./Google Chrome Framework.framework/Google Chrome Framework		      |	113048	./Google Chrome Framework.framework/Google Chrome Framework

And note this isn't Mac-specific. Here's the graph for Linux

https://chromeperf.appspot.com/report?sid=65d381ee52ca3e4a5a4ab2eb213dc7d0cf561050695ee068024828fc01ae62e9

We don't seem to monitor that plot for regressions (we should - filed  Issue 671403 )

Thanks for the help - it turns out my CL inadvertently included a generated js file which caused the 1.2MB file size jump in resources.pak. I'm working on a CL to fix this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8a3318ab2d4685e1edd8bfcda51f1d5d5c8d3e49

commit 8a3318ab2d4685e1edd8bfcda51f1d5d5c8d3e49
Author: chenwilliam <chenwilliam@chromium.org>
Date: Tue Dec 06 03:19:47 2016

DevTools: do not include unit_test_runner in resources.pak

My previous CL inadvertently caused a file size regression because unit_test_runner.js (~1.2MB)
was being included in resources.pak. This excludes unit_test_runner.js/html from being included
in the devtools_resources.grd and subsequently the resources.pak.

CL which caused file size regression:
https://codereview.chromium.org/2513423003

BUG= 670734 

Review-Url: https://codereview.chromium.org/2552963002
Cr-Commit-Position: refs/heads/master@{#436508}

[modify] https://crrev.com/8a3318ab2d4685e1edd8bfcda51f1d5d5c8d3e49/third_party/WebKit/Source/devtools/BUILD.gn

I checked the charts and the resources.pak file size regression has been fixed.