On 09/13/2012 04:20 PM, Alan Bateman wrote:
On 13/09/2012 09:10, Weijun Wang wrote:
http://cr.openjdk.java.net/~weijun/7198205/webrev.00/
The test hangs on this line
loadClass ("java.awt.Button", loader, true);
It looks like no AWT class can be loaded in the testing environment.
The fix uses "java.sql.Array" instead. I guess it should be a class
inside JRE but also not loaded yet.
It looks like this test was excluded because of proxy settings rather
than its usage of AWT (on Mac then using system proxy settings is
enabled by default and I think this causes problems on machines where
http connections to 127.0.0.1 are wrongly configured to go through the
proxy). So no issue with removing the dependency on AWT, I'm just not
sure that it was the reason it was excluded.
It was added thru
http://hg.openjdk.java.net/jdk8/tl/jdk/rev/9de97594f062
At first I thought this is the reason and the machine does shows an
environment variable http_proxy. However, I unset it and nothing
changed. After some println it seems the test only hangs at loading the
java.awt.Button class.
Also, aren't
java.net.useSystemProxies=false
http.nonProxyHosts=localhost|127.*|[::1]
default in net.properties? Shouldn't JRE on macosx honor them?
BTW: Just above it then looks like someone has put a security test in
the jdk_net section of the file, I think that should be moved to
jdk_security.
I put it there. It is a test in sun/security but the reason is net (No
PortUnreachableE on macosx). I thought putting it in the java_net
section would alert networking guys more. :)
Thanks
Max
-Alan.
