Initial checkin of a perf test suite for Indexed Database. This just has a

Initial checkin of a perf test suite for Indexed Database.  This just has a
tiny framework, a few tests of adding key-value pairs, and the mechanism to run
it all and report results.  This won't yet run on the bots.

If this looks good, I'll start adding tests to cover more of the API.

BUG=none
TEST=self


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=146173

[jsbell, 2012-06-21 00:04:15]

Looking good, mostly FYI notes.

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
File chrome/test/data/indexeddb/perf_shared.js (right):

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:3: window.IDBTransaction =
window.IDBTransaction || window.webkitIDBTransaction ||
Shouldn't need to reference IDBTransaction directly any more, so this can be
removed.

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:49: var os = objectStores;
What's this extra object reference for?

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:94: transaction.onerror = onError;
FYI, transaction.onerror sees errors bubbled up from individual requests which
were not handled/prevented. If the bubbling is not stopped here they hit the
database, and will eventually cause the transaction to abort. Therefore, it may
or may not be worth trapping onerror here. (i.e. may be extra log spam, etc). If
added for general logging, it could be done on the database connection's onerror
instead.

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
File chrome/test/data/indexeddb/perf_test.js (right):

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_test.js:25: os.put("key " + j, value);
This is measuring three distinct things:
* Time taken for the requests to be made (i.e. synchronous put calls)
* Time taken for the asynchronous requests to be processed
* Time for the commit to occur after the last request is processed (which could
be measured by the time delta between the last request.onsuccess and the
transaction.oncomplete)

We'll want to break those out eventually. (Many tests may interleave the first
two, but I suspect it will be common to have tests which are effectively three
different phases.)

[jsbell, 2012-06-21 00:04:25]

lgtm

[ericu, 2012-07-11 18:04:35]

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
File chrome/test/data/indexeddb/perf_shared.js (right):

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:3: window.IDBTransaction =
window.IDBTransaction || window.webkitIDBTransaction ||
On 2012/06/21 00:04:16, jsbell wrote:
> Shouldn't need to reference IDBTransaction directly any more, so this can be
> removed.

Removed.

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:49: var os = objectStores;
On 2012/06/21 00:04:16, jsbell wrote:
> What's this extra object reference for?

Removed; I was thinking I'd need it to get the closure right, but it turned out
not to be necessary.  I still need to look at things a few times to get closures
right in JS.

https://chromiumcodereview.appspot.com/10575043/diff/2001/chrome/test/data/in...
chrome/test/data/indexeddb/perf_shared.js:94: transaction.onerror = onError;
On 2012/06/21 00:04:16, jsbell wrote:
> FYI, transaction.onerror sees errors bubbled up from individual requests which
> were not handled/prevented. If the bubbling is not stopped here they hit the
> database, and will eventually cause the transaction to abort. Therefore, it
may
> or may not be worth trapping onerror here. (i.e. may be extra log spam, etc).
If
> added for general logging, it could be done on the database connection's
onerror
> instead.

I added a stopPropagation in onError.  In generally I'm going to try to err on
the side of verbosity on errors, especially for errors that should never happen
in the tests.  I also want to catch them as early as possible in the
propagation, so as to be able to localize them better, but I haven't put in that
bit yet.

[commit-bot: I haz the power, 2012-07-11 18:05:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ericu@chromium.org/10575043/10001

[commit-bot: I haz the power, 2012-07-11 18:05:39]

Presubmit check for 10575043-10001 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit Warnings **
License must match:
.*? Copyright \(c\) 2012 The Chromium Authors\. All rights reserved\.\n.*? Use
of this source code is governed by a BSD-style license that can be\n.*? found in
the LICENSE file\.(?: \*/)?\n
Found a bad license header in these files:
  chrome/test/data/indexeddb/perf_shared.js \
  chrome/test/data/indexeddb/perf_test.js

Presubmit checks took 3.9s to calculate.

[commit-bot: I haz the power, 2012-07-11 18:22:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ericu@chromium.org/10575043/11006

[commit-bot: I haz the power, 2012-07-11 19:35:57]

Change committed as 146173