Add more IDB perf tests, including cursor reads and log-style writes.

Add more IDB perf tests, including cursor reads and log-style writes.

BUG=137764
TEST=


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=149125

[ericu, 2012-07-26 23:53:13]

Alec, these are all tests you suggested, so you get this one.

[jsbell, 2012-07-27 00:25:46]

lgtm, minor comments

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
File chrome/test/data/indexeddb/perf_shared.js (right):

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:193: s = "" + s;
Although it should get optimized to the same thing, String(s) (NOTE: no "new"
operator) may be more readable.

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:195: while (s.length < width) {
Alternately: (Array(width + 1).join("0") + s).substr(-width)

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:305: var keyRange =
webkitIDBKeyRange.bound(
Should add at the top of the file:
window.IDBKeyRange = window.IDBKeyRange || window.webkitIDBKeyRange;
(Note that ONLY webkit prefixes this; other browsers didn't bother.)

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:324: values.push({key: cursor.key,
value: cursor.value});
Although the input/output stores here differ, having a cursor perf test that
does write to the same store that is being read from is a good idea. In the
current impl, writing to a store invalidates a per-cursor "prefetch" cache.

Huh, looking at the code, it looks like we reset ALL cursors, regardless of
store. And in many cases where we don't need to. So... um... yeah, this is a
good test. :)

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
File chrome/test/data/indexeddb/perf_test.js (right):

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_test.js:259: o.indexKeyPath = "field" + i; //
depends on getNFieldObjectValue
Maybe add function getNFieldName(n) { return "field" + n; } to wrap dependency?

[ericu, 2012-07-31 01:00:53]

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
File chrome/test/data/indexeddb/perf_shared.js (right):

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:193: s = "" + s;
On 2012/07/27 00:25:46, jsbell wrote:
> Although it should get optimized to the same thing, String(s) (NOTE: no "new"
> operator) may be more readable.

Done.

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:195: while (s.length < width) {
On 2012/07/27 00:25:46, jsbell wrote:
> Alternately: (Array(width + 1).join("0") + s).substr(-width)

I find that shorter, but a bit harder to read, so I'd prefer to leave this
as-is.

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:305: var keyRange =
webkitIDBKeyRange.bound(
On 2012/07/27 00:25:46, jsbell wrote:
> Should add at the top of the file:
> window.IDBKeyRange = window.IDBKeyRange || window.webkitIDBKeyRange;
> (Note that ONLY webkit prefixes this; other browsers didn't bother.)

Done.

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_shared.js:324: values.push({key: cursor.key,
value: cursor.value});
On 2012/07/27 00:25:46, jsbell wrote:
> Although the input/output stores here differ, having a cursor perf test that
> does write to the same store that is being read from is a good idea. In the
> current impl, writing to a store invalidates a per-cursor "prefetch" cache.
> 
> Huh, looking at the code, it looks like we reset ALL cursors, regardless of
> store. And in many cases where we don't need to. So... um... yeah, this is a
> good test. :)

Added a TODO to add a test that writes back to the same store.

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
File chrome/test/data/indexeddb/perf_test.js (right):

https://chromiumcodereview.appspot.com/10826038/diff/7/chrome/test/data/index...
chrome/test/data/indexeddb/perf_test.js:259: o.indexKeyPath = "field" + i; //
depends on getNFieldObjectValue
On 2012/07/27 00:25:46, jsbell wrote:
> Maybe add function getNFieldName(n) { return "field" + n; } to wrap
dependency?

Done.

[commit-bot: I haz the power, 2012-07-31 01:01:08]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ericu@chromium.org/10826038/6003

[commit-bot: I haz the power, 2012-07-31 02:37:59]

Change committed as 149125