The following six RMI regression tests have started failing this past weekend:

java/rmi/dgc/dgcImplInsulation/DGCImplInsulation.java
java/rmi/server/RMIClassLoader/checkPermission/CheckPermission.java
java/rmi/server/RMIClassLoader/spi/ContextInsulation.java
java/rmi/server/RMISocketFactory/setFailureHandlerPermission/SetFailureHandlerPermission.java
java/rmi/server/RemoteServer/setLogPermission/SetLogPermission.java
java/rmi/transport/checkAccept/CheckAccept.java

with compilation errors that look like this:

/files/JDK1.5/test/java/rmi/dgc/dgcImplInsulation/DGCImplInsulation.java:42: reference to CodeSource is ambiguous, both method CodeSource(java.net.URL,java.security.cert.Certificate[]) in java.security.CodeSource and method CodeSource(java.net.URL,java.security.CodeSigner[]) in java.security.CodeSource match
new ProtectionDomain(new CodeSource(null, null), perms) });
                                     ^
1 error
result: Failed. Compilation failed

All of the compilation errors are for code that attempts to invoke the following constructor of java.security.CodeSource:

public CodeSource(URL url, Certificate[] certs)

with a null "certs" argument. But the following constructor overload was recently added to CodeSource:

public CodeSource(URL url, CodeSigner[] signers)

which causes these constructor invocation attempts to be ambiguous at compile time. Casts need to be added to each such constructor invocation in the source code to clarify which constructor overload is intended.