Re: [cp-patches] Patch: RFC: ClassLoader and resolution

[Tom Tromey]

>>>>> "Archie" == Archie Cobbs <address@hidden> writes:

>> I wanted to post this for comment.  I couldn't easily test this inside
>> libgcj, due to an oddity of our implementation.  (I'll send a note
>> about that too, shortly.)
>> It seems to me that if you call loadClass("foo",false) and then
>> loadClass("foo",true), the second call will not actually resolve the
>> class, due to the early return in ClassLoader.loadClass().  This patch
>> fixes the apparent bug.

Archie> Looks fine to me.. I'm curious: did you actually trip over
Archie> this bug?  It seems like a pretty obscure case.

No, I just noticed it while reading the code.
You have to try pretty hard to even detect this situation.

Tom