Review request: 7120875 fix macos ipv6 issue and update multiple test scripts
Michael McMahon
michael.x.mcmahon at oracle.com
Wed Dec 14 12:57:31 PST 2011
On 14/12/11 19:54, Alan Bateman wrote:
> On 14/12/2011 19:39, Michael McMahon wrote:
>> Hi,
>>
>> This webrev fixes a bunch of test script issues in networking and
>> core-libs.
>> In most if not all cases, the change relates to recognising the OS
>> that the test
>> is running on.
>>
>> It also fixes an IPv6 issue, which requires Macos versions of a
>> couple of java.net
>> classes. These are the new files at the end of the webrev. The change
>> is to provide
>> a default sin6_scope_id when none is provided by the user.
>>
>> http://cr.openjdk.java.net/~michaelm/7120875/webrev.1/
>>
>> Thanks,
>> Michael.
> Michael - I haven't studied the IPv6 changes yet but I'm concerned
> that this creates a copy of java.net.MulticastSocket. Would some
> refactoring allow us to avoid this?
>
A check for "os.version" in the shared version of MulticastSocket would
work. There's
only a couple of lines of additional code required. But, then we don't
want os specific code
in the shared tree either. Don't know if you have other suggestions?
This is just to workaround an issue in Macosx
where (the Kame ipv6 stack is supposed to take care of default
scope_id's but doesn;t seem to be).
It's possible this issue may get resolved some time in the future, thus
eliminating the need for the code.
(I'll check this point with Mike anyway).
> In test/sun/net/www/protocol/jar/jarbug/run.sh then maybe it would
> make sense to combine SunOS, Linux and Darin into the one case. More
> generally then I guess most of these tests could be changed so that
> they default to PS=":" FS="/" when not on Windows or cygwin. Did you
> check the java code for the tests in these areas for any cases where
> they test based on os.name?
>
I'll let Kurchi deal with this one.
- Michael.
More information about the macosx-port-dev
mailing list