IndexedDB: Run cursor-prefetch webkit layout test as ui test.

IndexedDB: Run cursor-prefetch webkit layout test as ui test.

The test exercises a bug that is only triggered when run in multi-process chromium.  The patch in https://bugs.webkit.org/show_bug.cgi?id=75596 has to be committed and rolled before this change can be committed.

This ui test will eventually run all of the IDB layout tests.

BUG=108071
TEST=


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=117196

[dgrogan, 2012-01-05 04:41:06]

This obviously depends on the webkit patch[1] getting committed and rolled.

1. https://bugs.webkit.org/show_bug.cgi?id=75596

[jsbell, 2012-01-05 19:03:55]

Overall LGTM, just nits.

Have you tried with any other layout tests, e.g. some of the *-basics.html ones,
to see if work will be necessary to get them running this way?

http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
File content/browser/in_process_webkit/indexed_db_uitest.cc (right):

http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
content/browser/in_process_webkit/indexed_db_uitest.cc:41: // Runs each test in
an array of strings until it hits a NULL.
I see this pattern used in dom_storage_uitest.cc, but it looks like most of the
rest of Chromium uses an arraysize macro rather than a sentinel value.

http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
content/browser/in_process_webkit/indexed_db_uitest.cc:44: RunLayoutTest(*files,
kNoHttpPort);
Possibly pass the port into this function, so we don't get out of sync between
the call site which also passes it into InitializeForLayoutTest() - I can see us
changing this to kHttpPort at some future time as the security model around IDB
vs. file: evolves.

http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
content/browser/in_process_webkit/indexed_db_uitest.cc:53:
InitializeForLayoutTest(test_dir_, FilePath(), kNoHttpPort);
e.g.:
const int port = kNoHttpPort;
InitializeForLayoutTest(test_dir_, FilePath(), port);
AddJSTestResources();
RunTests(kLayoutTestFileNames, port);

[dgrogan, 2012-01-07 01:06:34]

On 2012/01/05 19:03:55, jsbell wrote:
> Overall LGTM, just nits.
> 
> Have you tried with any other layout tests, e.g. some of the *-basics.html
ones,
> to see if work will be necessary to get them running this way?

Yeah, a few.  Most work, one did not, probably because the output is too big. 
(I think layout-as-ui tests work by dumping the output to a cookie and then
comparing the cookie and golden.)  Still need to investigate.

>
http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
> File content/browser/in_process_webkit/indexed_db_uitest.cc (right):
> 
>
http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
> content/browser/in_process_webkit/indexed_db_uitest.cc:41: // Runs each test
in
> an array of strings until it hits a NULL.
> I see this pattern used in dom_storage_uitest.cc, but it looks like most of
the
> rest of Chromium uses an arraysize macro rather than a sentinel value.
> 
>
http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
> content/browser/in_process_webkit/indexed_db_uitest.cc:44:
RunLayoutTest(*files,
> kNoHttpPort);
> Possibly pass the port into this function, so we don't get out of sync between
> the call site which also passes it into InitializeForLayoutTest() - I can see
us
> changing this to kHttpPort at some future time as the security model around
IDB
> vs. file: evolves.

arraysize needs to know a compile-time constant parameter so this doesn't work
exactly, but I changed it to follow the general principle.

>
http://codereview.chromium.org/9108004/diff/1/content/browser/in_process_webk...
> content/browser/in_process_webkit/indexed_db_uitest.cc:53:
> InitializeForLayoutTest(test_dir_, FilePath(), kNoHttpPort);
> e.g.:
> const int port = kNoHttpPort;
> InitializeForLayoutTest(test_dir_, FilePath(), port);
> AddJSTestResources();
> RunTests(kLayoutTestFileNames, port);

[commit-bot: I haz the power, 2012-01-07 02:17:19]

No LGTM from valid reviewers yet.

[dgrogan, 2012-01-07 04:51:57]

Michael, could you review this?  Apparently Josh's LGTMs aren't good anymore. 
We'll have to nominate him as full committer.

[michaeln, 2012-01-10 21:22:41]

LGTM

http://codereview.chromium.org/9108004/diff/7002/content/browser/in_process_w...
File content/browser/in_process_webkit/indexed_db_uitest.cc (right):

http://codereview.chromium.org/9108004/diff/7002/content/browser/in_process_w...
content/browser/in_process_webkit/indexed_db_uitest.cc:30:
UILayoutTest::SetUp();
why is this override needed?

[dgrogan, 2012-01-11 08:27:18]

http://codereview.chromium.org/9108004/diff/7002/content/browser/in_process_w...
File content/browser/in_process_webkit/indexed_db_uitest.cc (right):

http://codereview.chromium.org/9108004/diff/7002/content/browser/in_process_w...
content/browser/in_process_webkit/indexed_db_uitest.cc:30:
UILayoutTest::SetUp();
On 2012/01/10 21:22:41, michaeln wrote:
> why is this override needed?

Whoops, it's a vestige.  Removed.

[commit-bot: I haz the power, 2012-01-11 08:30:09]

Failed to apply the patch.

[commit-bot: I haz the power, 2012-01-11 08:48:19]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dgrogan@chromium.org/9108004/18001

[commit-bot: I haz the power, 2012-01-11 10:03:11]

Change committed as 117196