[jsr294-modularity-eg] Static compilation

[Glyn Normington]

"Michal Cierniak" <cierniak at google.com> on 02/05/2007 17:52:32:

> On 5/2/07, Glyn Normington <glyn_normington at uk.ibm.com> wrote:
> > "Michal Cierniak" <cierniak at google.com> wrote on 01/05/2007 18:33:07:
> >
> >  > We haven't talked about such details but my
> >  > assumption was that it should be possible to dynamically load a 
class
> >  > that extends any non-final class including those classes that are 
not
> >  > exported.
> >
> > My expectation is different, as you may have guessed.
> 
> Glyn,
> 
> I think that your expectation is completely reasonable but I think
> that we would need more mechanisms to enforce this problem because the
> JSR-277 APIs as presented in the EDR allow you to arrive at the
> problem described by me.  Here's some pseudocode that shows my
> scenario (it's pseudo in the sense that I didn't look up carefully the
> APIs or tried to compile it, so that are probably some typos there):
> 
> class A {
>   // class is in a module M and is not exported
>   public void F() {
>      ClassLoader cl = A.class.getClassLoader();  // M's classloader
>      byte[] b = getBytes();  // bytes defining a class that subclasses A
>      Class b = cl.defineClass("B", b, 0, b.length);
>   }
>   byte[] getBytes() { ...}
> }
> 
> In this example, I use A's classloader to create class B which is a
> subclass of A.
> 
> Dealing with classloaders is always tricky, so I may be missing
> something here.  What do you think about my example?

Thanks for the example as it makes discussion easier.

I agree that more mechanisms would be required to enable static 
compilation, but I would have thought any static compilation system would 
constrain the contents of modules it was compiling pretty strongly to 
prevent the contents (classes in this example) changing at runtime. I 
probably see things differently to Rob Yates, but I wouldn't expect JSR 
294 to provide a "free ride" for a static compilation system - just give 
it the constructs to hang off. I'd expect static compilation to check its 
own constraints of modules it was compiling and also use some kind of 
sealing mechanism to preserve the integrity of the compiled output.

Glyn





Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 
741598. 
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU





-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/attachments/20070504/53995056/attachment.html