Add test that passes invalid arguments to socket methods and fix

Add test that passes invalid arguments to socket methods and fix
issues encountered.

The fuzzing is not as systematic as for some of the other
APIs. However, a lot of checking is going on in the Dart code so
forcing issues required a bit of work making objects act like lists
and integers.

Additionally, fix a memory leak in process implementation.

R=sgjesse@google.com
BUG=
TEST=

Committed: https://code.google.com/p/dart/source/detail?r=8088

[Søren Gjesse, 2012-05-30 07:07:15]

lgtm

https://chromiumcodereview.appspot.com/10441078/diff/1/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://chromiumcodereview.appspot.com/10441078/diff/1/runtime/bin/socket.cc#...
runtime/bin/socket.cc:233: const char* bind_address =
DartUtils::GetStringValue(bind_address_obj);
We could consider having GetStringValue work like GetInt64Value.

https://chromiumcodereview.appspot.com/10441078/diff/1/tests/standalone/stand...
File tests/standalone/standalone.status (right):

https://chromiumcodereview.appspot.com/10441078/diff/1/tests/standalone/stand...
tests/standalone/standalone.status:18: io/socket_invalid_arguments_test: Fail,
OK
I am still confused by the state "Fail, OK" :-|

[Mads Ager (google), 2012-05-30 07:35:11]

https://chromiumcodereview.appspot.com/10441078/diff/1/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://chromiumcodereview.appspot.com/10441078/diff/1/runtime/bin/socket.cc#...
runtime/bin/socket.cc:233: const char* bind_address =
DartUtils::GetStringValue(bind_address_obj);
On 2012/05/30 07:07:15, Søren Gjesse wrote:
> We could consider having GetStringValue work like GetInt64Value.

Yes, that could be done. It makes sense to have two version. One that asserts
and one that checks. I'll see if I can streamline that in an upcoming change.

https://chromiumcodereview.appspot.com/10441078/diff/1/tests/standalone/stand...
File tests/standalone/standalone.status (right):

https://chromiumcodereview.appspot.com/10441078/diff/1/tests/standalone/stand...
tests/standalone/standalone.status:18: io/socket_invalid_arguments_test: Fail,
OK
On 2012/05/30 07:07:15, Søren Gjesse wrote:
> I am still confused by the state "Fail, OK" :-|

It is failing and that is OK. :-)

What it means is that the test fails and that the failure is not something that
we want to fix. In the summary report (using the --report option to the test
scripts) of failing tests this is listed as a failure that we don't want to fix.