Mandy,
Looks fine to me. I agree that we should back it out for 7u40. I
would think we want to leave the change in jdk8 and continue the
investigation and resolving the JCK failures for jdk8. Is that what
you are thinking? If so, we don't need to back it out from jdk8.
I was hoping to back out the fix for both jdk7u40 and jdk8. The setup to
reproduce is quite simple (modify package list in java.security) and I
think it's not necessary to have JCK failures in jdk8 for the short to
medium term as a result.
regards,
Sean.
On 12/07/2013 22:37, Mandy Chung wrote:
On 7/12/2013 9:14 PM, Seán Coffey wrote:
The recent 8000450 has caused issue for JCK testing and some RMI/JMX
testing also.The CORBA package interaction between org.omg,
com.sun.corba.se.spi, javax.rmi and other packages into
com.sun.corba.se.impl classes needs better analysis and won't be
complete for 7u40.
It's rare to have security manager installed with CORBA app but we
need to cater for it. The JCK testsuite tests this scenario but it
doesn't have sufficient privileges in the stack when launching the
CORBA server during test setup phase. None the less, it's something
the JDK code should handle. The structure of the CORBA packages and
users of it needs to be better scoped out. For now, I propose
reversing the change for both jdk8 and jdk7u40.
webrev : http://cr.openjdk.java.net/~coffeys/webrev.8000450.7u40/webrev/
Looks fine to me. I agree that we should back it out for 7u40. I
would think we want to leave the change in jdk8 and continue the
investigation and resolving the JCK failures for jdk8. Is that what
you are thinking? If so, we don't need to back it out from jdk8.
Mandy
regards,
Sean.
