Re: [BC ABI] For entertainment purposes only: gcj-JIT

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Bryce" == Bryce McKinlay <mckinlay@redhat.com> writes:

Bryce> See this thread for the discussion the last time this issue came up:
Bryce> http://gcc.gnu.org/ml/java/2004-07/msg00219.html

Thanks, I'll re-read.

Bryce> - It doesn't seem very likely that any sane code actually
Bryce> relies on this

There was code in Eclipse relying on this.  As I recall, one of the
plugins in the startup sequence relied on lazy resolution in order to
install a ClassLoader that would be used to load some of the classes
to which it later referred.

I doubt they were aware of this though.  And I agree, it is more than
arguably an Eclipse bug, and that code doesn't really qualify as
"sane".

Bryce> - Any code that already runs on GIJ will be fine with the current BC
Bryce>   approach, because GIJ is actually not yet compliant with the spec,
Bryce>   it is currently too eager at throwing exceptions for failed
Bryce>   resolution.

Could you file a PR for this (if there isn't one already)?

For the proximate problem, see the patch I just committed.  Simply
preparing the target class solve the problem Andrew was referring to.

Tom