Added TodoMVC startup benchmarks.

Added TodoMVC startup benchmarks.

R=jmesserly@google.com

Committed: https://code.google.com/p/dart/source/detail?r=34561

[Emily Fortuna, 2014-03-19 17:59:36]

Oh boy! It's another enormous CL! 
Here's the lowdown: 

1) samples/third_party/todomvc_performance is mostly a copy of
samples/third_party/toodmvc, with the following exceptions:
- samples/third_party/todomvc_performance/js_todomvc has been added as a pure JS
implementation of todo MVC. It is a copy of this github repo:
https://github.com/Polymer/todomvc
- modified samples/third_party/todomvc_performance/pubspec.yaml
- added samples/third_party/todomvc_performance/web/performance.dart and
samples/third_party/todomvc_performance/web/startup-performance.html

2) Added on to tools/testing/perf_testing/run_perf_tests.py to add this to our
automated performance tests.

[Jennifer Messerly, 2014-03-19 19:27:57]

Note: I didn't review the JS code from Bower. Assuming there are no changes ;-)

most of my comments are minor/optional, the only thing that jumped out was
customElementsReady vs Polymer.onReady -- could we always use the latter?

https://codereview.chromium.org/204733004/diff/10001/.gitignore
File .gitignore (right):

https://codereview.chromium.org/204733004/diff/10001/.gitignore#newcode68
.gitignore:68: pkg/shadow_dom/tool/node_modules
I just noticed this line is old. If you don't mind, maybe remove? :)

https://codereview.chromium.org/204733004/diff/10001/.gitignore#newcode72
.gitignore:72: samples/third_party/dromaeo/build
fyi -- you can also add a .gitignore inside 

samples/third_party/dromaeo

but this works too.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/README.md (right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/README.md:3: > Build structured,
encapsulated, client-side web apps with Dart and web components.
remove or change this README?

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/pubspec.yaml (right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/pubspec.yaml:6: browser_controller:
">=0.00.1-dev <0.0.3"
most unstable pkg ever! :)

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/web/performance.dart (right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:1: // Copyright (c)
2013, the Dart project authors.  Please see the AUTHORS file
not a big deal but could be 2014 :)

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:5: library
todomvc.test.markdone_test;
change library name?

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:27:
Polymer.onReady.then((_) {
fyi -- I think this event is always more reliable as far as Polymer.dart
startup.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/web/startup-performance.html
(right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:4:
Copyright (c) 2013, the Dart project authors.  Please see the AUTHORS file
2014?

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:16: var
startTime = new Date().getTime();
another possibly interesting start time is the various ones on
window.performance.timing
https://developer.mozilla.org/en-US/docs/Navigation_timing

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:24: // a
fallback case.
any reason not to use Polymer.onReady also?

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:27: var
endTime = e.data;
another idea here, this code could be moved into Dart either using the
performance timing API or JS interop to fetch the start time, e.g.

<script>window.jsStartTime = new Date().getTime();</script>

// In Dart:
var startTime = js.context['jsStartTime'];
// ... append the printout ...

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
File tools/testing/perf_testing/run_perf_tests.py (right):

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
tools/testing/perf_testing/run_perf_tests.py:741: 'dart-sdk', 'bin', 'pub'),
'build', '--mode=debug'])
should this be debug? Might be worth a comment either way :)
we don't do anything especially slow in Polymer.dart for mode=debug, but we
might start someday :D

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
tools/testing/perf_testing/run_perf_tests.py:854: dromaeo_path =
os.path.join('samples', 'third_party', 'dromaeo')
future TODO: maybe we should move both of these out of "samples"? I seem to
recall a request a while back to move dromaeo if possible. Maybe under
tests/perf/... or something along those lines.

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
tools/testing/perf_testing/run_perf_tests.py:858: 'todomvc_performance'))
fyi -- I was thinking of moving TodoMVC out of third_party
(https://code.google.com/p/dart/issues/detail?id=17599), not sure if that
impacts the path here. Then again since it includes the JS code maybe
"third_party" still makes sense. Polymer-JS uses the same license as Dart but I
think the transitive closure includes things like "flatiron"

[Emily Fortuna, 2014-03-24 19:45:54]

ptal, please. :-)

https://codereview.chromium.org/204733004/diff/10001/.gitignore
File .gitignore (right):

https://codereview.chromium.org/204733004/diff/10001/.gitignore#newcode68
.gitignore:68: pkg/shadow_dom/tool/node_modules
On 2014/03/19 19:27:57, John Messerly wrote:
> I just noticed this line is old. If you don't mind, maybe remove? :)

removed!

https://codereview.chromium.org/204733004/diff/10001/.gitignore#newcode72
.gitignore:72: samples/third_party/dromaeo/build
On 2014/03/19 19:27:57, John Messerly wrote:
> fyi -- you can also add a .gitignore inside 
> 
> samples/third_party/dromaeo
> 
> but this works too.

added a .gitignore up in that dir.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/web/performance.dart (right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:1: // Copyright (c)
2013, the Dart project authors.  Please see the AUTHORS file
On 2014/03/19 19:27:57, John Messerly wrote:
> not a big deal but could be 2014 :)

Done.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:5: library
todomvc.test.markdone_test;
On 2014/03/19 19:27:57, John Messerly wrote:
> change library name?

Done.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/performance.dart:27:
Polymer.onReady.then((_) {
On 2014/03/19 19:27:57, John Messerly wrote:
> fyi -- I think this event is always more reliable as far as Polymer.dart
> startup.

Done.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
File samples/third_party/todomvc_performance/web/startup-performance.html
(right):

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:4:
Copyright (c) 2013, the Dart project authors.  Please see the AUTHORS file
On 2014/03/19 19:27:57, John Messerly wrote:
> 2014?

Done.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:16: var
startTime = new Date().getTime();
On 2014/03/19 19:27:57, John Messerly wrote:
> another possibly interesting start time is the various ones on
> window.performance.timing
> https://developer.mozilla.org/en-US/docs/Navigation_timing

Hm, would performance.timing.navigationStart be a better measure starting point?

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:24: // a
fallback case.
On 2014/03/19 19:27:57, John Messerly wrote:
> any reason not to use Polymer.onReady also?

Done.

https://codereview.chromium.org/204733004/diff/10001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:27: var
endTime = e.data;
On 2014/03/19 19:27:57, John Messerly wrote:
> another idea here, this code could be moved into Dart either using the
> performance timing API or JS interop to fetch the start time, e.g.
> 
> <script>window.jsStartTime = new Date().getTime();</script>
> 
> // In Dart:
> var startTime = js.context['jsStartTime'];
> // ... append the printout ...
> 

I wrote it this way because I want to actually time the full startup time, and
Dart doesn't always start executing until after the other data is loaded. By
doing the initial call in JS at the top of the file, I can start timing a little
earlier.

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
File tools/testing/perf_testing/run_perf_tests.py (right):

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
tools/testing/perf_testing/run_perf_tests.py:741: 'dart-sdk', 'bin', 'pub'),
'build', '--mode=debug'])
On 2014/03/19 19:27:57, John Messerly wrote:
> should this be debug? Might be worth a comment either way :)
> we don't do anything especially slow in Polymer.dart for mode=debug, but we
> might start someday :D

Yes, this is debug on purpose. Siggi and I discussed this at length last week. I
can fill you in in person for the complete explanation, but I'll also add a
little comment here, too.

https://codereview.chromium.org/204733004/diff/10001/tools/testing/perf_testi...
tools/testing/perf_testing/run_perf_tests.py:854: dromaeo_path =
os.path.join('samples', 'third_party', 'dromaeo')
On 2014/03/19 19:27:57, John Messerly wrote:
> future TODO: maybe we should move both of these out of "samples"? I seem to
> recall a request a while back to move dromaeo if possible. Maybe under
> tests/perf/... or something along those lines.

I'm going to leave this as is for now because of the License stuff, but I'm open
to moving it in the future.

[Emily Fortuna, 2014-03-28 19:22:46]

Hi John,
Just a reminder, please take a look at this when you're able. I can't check in
any perf testing updates until it's in. :-)

[Jennifer Messerly, 2014-03-28 19:51:06]

lgtm

https://codereview.chromium.org/204733004/diff/60001/samples/third_party/todo...
File samples/third_party/todomvc_performance/pubspec.yaml (right):

https://codereview.chromium.org/204733004/diff/60001/samples/third_party/todo...
samples/third_party/todomvc_performance/pubspec.yaml:6: browser_controller:
#">=0.00.1-dev <0.0.3"
did you mean to change this back before checkin?

https://codereview.chromium.org/204733004/diff/70001/samples/third_party/todo...
File samples/third_party/todomvc_performance/web/startup-performance.html
(right):

https://codereview.chromium.org/204733004/diff/70001/samples/third_party/todo...
samples/third_party/todomvc_performance/web/startup-performance.html:17: //
performance.timing.navigationStart or performance.timing.domLoading?
yeah, the nice thing about using these times is you won't need this script tag
here at all :)
all the work could be done in the Dart main() method. but yeah, it's just a
small tweak, not super important.

[Emily Fortuna, 2014-03-31 17:09:38]

Committed patchset #5 manually as r34561 (presubmit successful).