[rfc][icedtea-web] DeploymentPropertiesAreExposed reproducer fix

[Omair Majid]

On 09/16/2013 10:02 AM, Jiri Vanek wrote:
> Yes it is interesting. From this point of view (and as Jacob picked up
> easily fake-able) both
>>> +    public final static String USER_CONFIG_HOME;
>>>       public final static String USER_CACHE_HOME;
> and also all other for-testing-package-private should become private and
> in tests used via reflection only...
> 
> Any opinion?

Please read this carefully:
http://docs.oracle.com/javase/specs/jvms/se7/html/jvms-5.html#jvms-5.4.4

In the normal netx runtime, Defaults will be loaded by the standard java
classloader (sun.misc.Launcher.AppClassLoader) but the jnlp/applet will
be loaded by net.sourceforge.jnlp.runtime.JNLPClassLoader. Even if an
applet claims it was the net.source.jnlp.* package, the runtime packages
are different. A package-private Defaults.USER_CONFIG_HOME will not be
visible to an applet.

That said, I wont mind if reproducers (not unit tests) did not poke into
implementation classes at all, and relied on external interfaces.

Thanks,
Omair

-- 
PGP Key: 66484681 (http://pgp.mit.edu/)
Fingerprint = F072 555B 0A17 3957 4E95  0056 F286 F14F 6648 4681