Porting ppapi_tests framework to postMessage.

Porting ppapi_tests framework to postMessage.

Some tests still rely on scripting, so we changed to using InstancePrivate (since scripting will disappear from Instance soon). Also use conditional compilation so that if compiled as untrusted with NaCl, the tests use Instance instead of InstancePrivate. This means that tests which rely on scripting aren't runnable in NaCl.

I also added a gyp option: pepper_scripting. The default is that scripting is on in this CL, but it will make it easy to turn it off in local builds. Soon we'll switch the default to no scripting, and soon after we can remove the option entirely.

BUG=82606
TEST=these tests

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=91859

[sehr (please use chromium), 2011-07-07 00:56:13]

On 2011/07/06 17:32:13, dmichael1 wrote:

LGTM.

[cstefansen, 2011-07-07 18:03:09]

Minor questions/comments.

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.h
File ppapi/tests/test_case.h (right):

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.h#newc...
ppapi/tests/test_case.h:86: pp::VarPrivate test_object_;
#if  !(defined __native_client__) ?

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.html
File ppapi/tests/test_case.html (right):

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.html#n...
ppapi/tests/test_case.html:87: var delim1 = message_data.indexOf(delim_str);
Consider using split() - it would simplify this code.

[dmichael (off chromium), 2011-07-07 21:17:59]

Comments addressed, and I think this fixes the out-of-process tests (turns out
PPB_Messaging proxy wasn't really working).

PTAL

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.h
File ppapi/tests/test_case.h (right):

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.h#newc...
ppapi/tests/test_case.h:86: pp::VarPrivate test_object_;
On 2011/07/07 18:03:09, cstefansen wrote:
> #if  !(defined __native_client__) ?
Good catch, done.

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.html
File ppapi/tests/test_case.html (right):

http://codereview.chromium.org/7312008/diff/4001/ppapi/tests/test_case.html#n...
ppapi/tests/test_case.html:87: var delim1 = message_data.indexOf(delim_str);
On 2011/07/07 18:03:09, cstefansen wrote:
> Consider using split() - it would simplify this code.
I wanted to do that, but it doesn't quite have the behavior I need. I want to
ensure that the 'message_content' part stays in-tact if there are any colons in
it. split also doesn't have the right semantics for my needs if I provide the
optional split limit.

[brettw, 2011-07-07 21:32:22]

LGTM

[cstefansen, 2011-07-07 21:41:12]

LGTM provided you verify that tests/ppapi_tests work on the NaCl side with this
patched into Chrome and then DEPSed into NaCl. (So we avoid a later breakage
when DEPS are rolled.)

[dmichael (off chromium), 2011-07-08 16:29:00]

I had to disable the Scrollbar test on Mac for now. The scrollbar itself is not
broken, just the test; it must be some weird timing issue.
crbug.com/88754
I wanted to unblock sehr's scripting removal, since it's going to take me a
while to get a chromium build on mac again to figure out the problem.