partial support for Safari testing

With this CL, browser tests of safari work when connecting to a remote
webdriver.  Meaning:
  1. start "java -jar selenium-standalone*.jar"
  2. run "ant brtest -Dtest.browser=safari -Dtest.webdriver.url=http://127.0.0.1:4444/wd/hub"

the main fix in this CL is that window.open doesn't work in the
Safari webdriver.

There's also code to start a local SafariDriver, but that doesn't work
yet. (There's a weird protocol failure when trying to send a command
to the webdriver Safari extension, and I haven't figured out why yet.)

[kpreid2, 2013-04-29 23:34:28]

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
File tests/com/google/caja/plugin/WebDriverHandle.java (right):

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:75: // ignore
Please re-add the UnsupportedCommandException case to ignore, and make this
broader case log the exception, so early symptoms of new problems aren't
muffled.

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:87: // window.open silently
fails on Safari
This comment may be true but does not seem to explain anything about the
adjacent code. Perhaps move this to a doc comment like "window.open silently
fails on Safari, so this method may not create a new distinct window."

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:123: // closed our only
window, and we'll need a new webdriver session.
I think it would read better if this information were in the catch block.

[felix8a, 2013-04-29 23:57:37]

With this CL, browser tests of safari work when connecting to a remote
webdriver.  Meaning:
  1. start "java -jar selenium-standalone*.jar"
  2. run "ant brtest -Dtest.browser=safari
-Dtest.webdriver.url=http://127.0.0.1:4444/wd/hub"

the main fix in this CL is that window.open doesn't work in the
Safari webdriver.

There's also code to start a local SafariDriver, but that doesn't work
yet. (There's a weird protocol failure when trying to send a command
to the webdriver Safari extension, and I haven't figured out why yet.)

[felix8a, 2013-04-29 23:58:03]

new snapshot

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
File tests/com/google/caja/plugin/WebDriverHandle.java (right):

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:75: // ignore
the Safari driver doesn't support the command and sends a WebDriverException
instead of UnsupportedCommandException. I think only the IE driver sends an
UnsupportedCommandException, which is a subclass of WebDriverException. I'll add
a noisy useless error message, but I don't see any value in catching the
more-specific exception separately.

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:87: // window.open silently
fails on Safari
On 2013/04/29 23:34:28, kpreid2 wrote:
> This comment may be true but does not seem to explain anything about the
> adjacent code. Perhaps move this to a doc comment like "window.open silently
> fails on Safari, so this method may not create a new distinct window."

Done.

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:123: // closed our only
window, and we'll need a new webdriver session.
On 2013/04/29 23:34:28, kpreid2 wrote:
> I think it would read better if this information were in the catch block.

Done.

[kpreid2, 2013-04-30 00:11:23]

LGTM

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
File tests/com/google/caja/plugin/WebDriverHandle.java (right):

https://codereview.appspot.com/9022043/diff/1/tests/com/google/caja/plugin/We...
tests/com/google/caja/plugin/WebDriverHandle.java:75: // ignore
On 2013/04/29 23:58:03, felix8a wrote:
> the Safari driver doesn't support the command and sends a WebDriverException
> instead of UnsupportedCommandException. I think only the IE driver sends an
> UnsupportedCommandException, which is a subclass of WebDriverException.

Sounds like a bug to file upstream.

https://codereview.appspot.com/9022043/diff/5001/tests/com/google/caja/plugin...
File tests/com/google/caja/plugin/WebDriverHandle.java (right):

https://codereview.appspot.com/9022043/diff/5001/tests/com/google/caja/plugin...
tests/com/google/caja/plugin/WebDriverHandle.java:89: // in the base window and
open a new session for the next test.
This isn't a doc comment as I requested, but it's adequately explanatory. OK.

[felix8a, 2013-04-30 00:15:44]

@r5384