All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=638624

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgztzqogoM


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

mac

Hi tsergeant@, do you know whether r412420 could have possibly affected Mac OS X Chrome binary size (GoogleChrome.app)?

Yup. As it says in the final paragraph of the description, that CL causes a size regression in resources.pak, but a follow-up patch (which I plan to land today) reduces the size delta significantly.

Awesome, thanks! Could you also add this bug number to that CL?

After it lands, we can also use these graphs to verify the change in size :-)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069

commit ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069
Author: tsergeant <tsergeant@chromium.org>
Date: Fri Aug 19 06:39:36 2016

MD WebUI: Uglify vulcanized javascript bundles to remove comments/whitespace

We run uglify in 'beautify' mode so that output is still human-readable,
but comments and whitespace are removed. This reduces the size of
resources.pak by 330K.

BUG= 629406 , 638624 
NOPRESUBMIT=true # crisper.js
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2257723002
Cr-Commit-Position: refs/heads/master@{#413070}

[modify] https://crrev.com/ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069/chrome/browser/resources/md_downloads/crisper.js
[modify] https://crrev.com/ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069/chrome/browser/resources/md_history/app.crisper.js
[modify] https://crrev.com/ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069/chrome/browser/resources/vulcanize.py
[modify] https://crrev.com/ce13eef039ce8ac44ae999ce62fd1bc3e0f0d069/docs/vulcanize.md

I tried to verify with an official windows build how much of a difference this makes, but just got nonsense results. I would expect that more than half of the increase in r412420 would be removed in r413070, but who knows what will happen when everything is compressed.

from what I can tell, the 400k you added was counter-balanced by the 400k you removed with uglify.  AND none of this mattered to the installer size, just the uncompressed, on-disk size (at least on mac)

https://chromeperf.appspot.com/report?sid=0b68579269a9a5a4402ed6ac605e79dc9b793a1ae695f94b6ec0b94117e9fede

The decrease in size can also be seen from the graphs, so that was as expected.

Anyway, I think in general binary size increase is acceptable as a trade-off, the main thing we'd be concerned about is if somebody accidentally introduced a large size increase without being aware of it.