I was trying out the --permit-illegal-access flag with a simple class that uses
an encapsulated type from the JDK by (non-reflectively) instantiating it in the
main method:
$ java Test
Exception in thread "main" java.lang.IllegalAccessError: class Test (in unnamed
module @0x7b3300e5) cannot access class sun.security.x509.X500Name (in module
java.base) because module java.base does not export sun.security.x509 to
unnamed module @0x7b3300e5
at Test.main(Test.java:5)
$ java --permit-illegal-access Test
WARNING: --permit-illegal-access will be removed in the next major release
What I expected to see was a warning message about sun.security.x509.X500Name
during the second invocation of java. Then, after reading [0] again, I realized
the warnings only pertain to reflective access.
What's the idea behind this distinction? The Test class fails just as well with
an IllegalAccessException without the --permit-illegal-access flag. Only, I
don't get to see the problem when running with --permit-illegal-access unless
the access is done reflectively. Is it because non-reflective uses can be found
statically by jdeps, and are not 'worthy' of a warning? Just looking for the
rationele here. It does mean fixing all warnings logged by
--permit-illegal-access is not a guarantee there won't be
IllegalAccessExceptions anymore afterwards.
Sander
[0] http://mail.openjdk.java.net/pipermail/jigsaw-dev/2017-March/011763.html
