Create a run_multipage_benchmarks entry point for all benchmarks.

Create a run_multipage_benchmarks entry point for all benchmarks.

Usage is:
./run_multipage_benchmarks [options] <benchmark> <page_set>

Where <benchmark> is the name of the file containing the benchmarks to run.

This patch doesn't handle benchmark discovery which will find all *_benchmark.py
files and run all classes that subclass PageTest.

BUG=None
TEST=None

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=162450

[tonyg, 2012-10-16 23:12:58]

Like we talked about, this is just a step towards automatic discovery to get the
command syntax nailed down. It still needs the actual discovery mechanism.
Please let me know what you think of this factoring.

[dtu, 2012-10-16 23:53:43]

nice! cleaner than before. lgtm

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
File tools/perf/perf_tools/multipage_benchmark_runner.py (right):

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:46:
sys.stderr.write('Available benchmarks:\n%s\n\n' % ',\n'.join(
print >> sys.stderr, 'Available bench...'

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:60: if possible_browser ==
None:
if not possible_browser:

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:70: # purpose of outputting
the perf results.
Not sure about the validity of this assumption. What if you're comparing two
builds?

[tonyg, 2012-10-17 00:13:44]

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
File tools/perf/perf_tools/multipage_benchmark_runner.py (right):

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:46:
sys.stderr.write('Available benchmarks:\n%s\n\n' % ',\n'.join(
On 2012/10/16 23:53:44, Dave Tu wrote:
> print >> sys.stderr, 'Available bench...'

Done.

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:60: if possible_browser ==
None:
On 2012/10/16 23:53:44, Dave Tu wrote:
> if not possible_browser:

Done.

https://chromiumcodereview.appspot.com/11199003/diff/1/tools/perf/perf_tools/...
tools/perf/perf_tools/multipage_benchmark_runner.py:70: # purpose of outputting
the perf results.
On 2012/10/16 23:53:44, Dave Tu wrote:
> Not sure about the validity of this assumption. What if you're comparing two
> builds?

It just changes the string that is printed in the results summary, so it doesn't
really hurt if this assumption breaks down. Anyway, this method was all just
copied from multi_page_benchmark.py, so it's not really the point of the patch.