You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Nathan Beyer <nd...@apache.org> on 2009/03/28 05:11:38 UTC
Re: [classlib][luni] - SO_REUSEADDR's default value of ServerSocket
should be true on Linux (HARMONY-6090)
Are both patches required? I applied the latest patch, but the class
PlainServerSocketImpl seems to be missing.
-Nathan
On Fri, Mar 27, 2009 at 3:40 PM, Tim Ellison <t....@gmail.com> wrote:
> I'm looking for another committer to review this patch with a view to
> fixing a regression from M8. This is (part of) the cause of the RMI
> test failure blocker.
>
> Briefly,
>
> Creating a new server socket has different default values for the
> SO_REUSEADDR flag between Windows and Linux on the RI, but not Harmony.
> This patch makes us match the RI behavior once again.
>
> Thanks,
> Tim
>
>
> Tim Ellison (JIRA) wrote:
>>
>> Tim Ellison updated HARMONY-6090:
>> ---------------------------------
>>
>> Attachment: harmony-6090.patch
>>
>> Alternative patch, that uses a new native to distinguish the server socket case, so the so_reuseaddr can be set as default for Linux.
>>
>>
>>> [classlib][luni] - the SO_REUSEADDR's default value of ServerSocket should be true on Linux
>>> -------------------------------------------------------------------------------------------
>>>
>>> Key: HARMONY-6090
>>> URL: https://issues.apache.org/jira/browse/HARMONY-6090
>>> Project: Harmony
>>> Issue Type: Bug
>>> Components: Classlib
>>> Affects Versions: 5.0M8
>>> Reporter: Regis Xu
>>> Assignee: Sean Qiu
>>> Fix For: 5.0M9
>>>
>>> Attachments: HARMONY-6090.diff, harmony-6090.patch
>>>
>>>
>>> In Harmony, the SO_REUSEADDR's default value of ServerSocket is false on Linux, while RI is true.
>>> according to spec, this default value is undefined, but it's better following RI to get better compatibility
>>
>
Re: [classlib][luni] - SO_REUSEADDR's default value of ServerSocket
should be true on Linux (HARMONY-6090)
Posted by Tim Ellison <t....@gmail.com>.
Nathan Beyer wrote:
> Are both patches required? I applied the latest patch, but the class
> PlainServerSocketImpl seems to be missing.
No, the PlainServerSocketImpl is a new class that I forgot to svn add
before creating the patch.
I've attached a new patch to the JIRA [1] called "harmony-6090-v2.patch"
which is the full, stand-alone patch being proposed to resolve this issue.
Thanks for taking a look.
[1] https://issues.apache.org/jira/browse/HARMONY-6090
Regards,
Tim