So long bootclasspath!
stef at epardaud.fr
Tue Dec 1 15:53:52 UTC 2015
On 30/11/15 17:14, Alan Bateman wrote:
> I assume you mean it can't access the types in
> sun.reflection.annotation rather than can't see them. In any case, it
> is a shame that there are types here that are used in annotation
> serialization. You can workaround it by using -XaddExports, documented
> in JEP 261.
Yes indeed, but that is viral, right? Every user will have to provide
this for every java invocation, and they aggregate too, because I guess
there will be other special cases too.
That doesn't sound very satisfying to me. After I've renamed my forked
modules that's the only issue remaining in our own code that uses JDK
internals, and then again, only because javac uses it and I don't know
how to rewrite it without it.
Short of making java.base export that package (which I understand you
guys don't want to do, and I agree with that choice), then I don't quite
see how to make it work. Unless I can rewrite the javac part that uses
this, but I guess I'd have to ask a javac guy if they think it's
possible. I don't really know this part of the code.
Otherwise I've also found module access errors in JBoss Modules (that
I'm sure we can fix) and Ant's xmlfilter (which is abandonned but others
have forked it with the fix). Nothing impossible to fix.
More information about the jigsaw-dev