On 16/12/2011 10:36, Michael McMahon wrote:
Updated webrev after Alan's comments.
http://cr.openjdk.java.net/~michaelm/7120875/webrev.2/
This is better but would be nice if we could avoid special casing MacOSX
in java.net.MulticastSocket. Given that we may have to choose the
default interface on other platforms when what would you think of
MulticastSocket checking for default interface on all platforms? (could
return null on platforms where it's not required). Same thing for
java.net.PlainSocketImpl.
The changes to the tests mostly look okay to me.
There are a couple of tests where additional debugging messages have
been added, I don't know if they are meant to be included, examples:
test/java/net/DatagramSocket/B6411513.java
java/net/DatagramSocket/SetDatagramSocketImplFactory/ADatagramSocket.java.
test/java/net/DatagramSocket/Send12k.java - now the send size is a local
then it might be better to name is "size" rather than SEND_SIZE.
There are a couple of tests the Darwin case has been combined with the
Linux case, examples:
test/java/net/URLClassLoader/B5077773.sh
test/java/net/DatagramSocket/SetDatagramSocketImplFactory/ADatagramSocket.sh
test/java/net/URLConnection/6212146/test.sh
No big deal but might be worth combining Linux too as further
maintainers might wonder why Linux is separate to SunOS and Darwin.
I exchanged a few mails with Kurchi about java tests as I think there
are a couple of tests that need to be updated, examples:
test/java/net/ipv6tests/Tests.java
test/java/net/DatagramSocket/SendDatagramToBadAddress.java
test/java/util/zip/ZipFile/ManyZipFiles.java
test/java/lang/ProcessBuilder/Basic.java
No problem if they are skipped in this round and I was just sending a
reminder that java tests may need to be updated too (even if they pass).
-Alan.
