Approved for jdk7u40-dev.

On Jul 31, 2013, at 1:44 PM, Alan Bateman <[email protected]> wrote:
> 
> A few weeks ago we pushed a change that "disables" the JDK private method 
> sun.reflect.Reflection.getCallerClass(int). Those that have been following 
> the effort to check the usages of caller sensitive methods in the JDK will be 
> familiar with the background to this.
> 
> At the time it was known that at least some developers have been using this 
> method directly and the solution allowed for those cases by providing a 
> system property to "re-enable" the method. Since then it has become clear 
> that there are a more usages that originally expected and disabling it by 
> default in 7u40 is going to cause significant pain and noise.
> 
> I would like to propose that we change 7u40 so that this method is not 
> disabled by default. This should not be interpreted to mean that we support 
> direct usages of JDK-private sun.* APIs of course. Note that I am not 
> proposing at this time to do anything for JDK 8, instead we need to take the 
> time to look at the use-cases in more detail to see what if there are other 
> changes or APIs that need to be introduced.
> 
> So for 7u40, I propose is to leave the system property in place and simply 
> change its default value. This allows for testing with the method disabled to 
> see if there are dependencies. It also means the changes (at this late time 
> in the release) are minimal. The webrev with the changes is here:
> 
> http://cr.openjdk.java.net/~alanb/8021946/webrev/
> 
> This change requires a jdk7u reviewer as it is not a backport. It also 
> requires one of the jdk7u maintainers to approve this change for the 7u40 
> stabilization forest.
> 
> -Alan.

Reply via email to