8218559: Reimplement the Legacy Socket API

Michael McMahon michael.x.mcmahon at oracle.com
Tue May 7 15:44:34 UTC 2019


Hi Alan,

What's the purpose of the change to the UdpTest?

- Michael.

On 01/05/2019, 14:10, Alan Bateman wrote:
>
> JEP 353 [1] is now a candidate and I would like to get the CSR [2] 
> finalized and the changes reviewed so that it can be targeted.
>
> The webrev with the changes is here:
>    http://cr.openjdk.java.net/~alanb/8221481/1/webrev/index.html
>
> The new implementation is one source file; the only other code change 
> is to SocketImpl.createPlatformSocketImpl to select the implementation 
> based on the system property jdk.net.usePlainSocketImpl. The reason 
> the patch isn't too big is that all the "enabling changes" are already 
> in jdk/jdk.
>
> We added additional tests a few weeks ago via JDK-8221259 [3] to test 
> long standing behavior. The webrev adds a few additional tests, the 
> main new one is to test that the SocketImpl is reliable in the fact of 
> usages that are not specified by SocketImpl. A selection of existing 
> tests are also updated to run with both the old and new implementation 
> - the motive of this is to ensure that the old implementation doesn't 
> bit rot before it is removed.
>
> Once the changes are in a promoted build then we'll need help testing. 
> The best outcome is that nobody notices :-)  Rory O'Donnell, with his 
> OpenJDK quality outreach hat on, has agreed to contact the projects 
> tracked by that program to test the EA builds when they are available.
>
> -Alan
>
> [1] https://openjdk.java.net/jeps/353
> [2] https://bugs.openjdk.java.net/browse/JDK-8222028
> [3] https://bugs.openjdk.java.net/browse/JDK-8221259


More information about the net-dev mailing list