On Jun 6, 2011, at 6:25 AM, Richard Ellis wrote:

> I'm sorry if I wasn't clear in making my point, which I believe has now 
> been clarified by Tim. It was in fact that the specific Aries proxy 
> methods under discussion were marked final for a reason other than 
> performance and that reason was so that they could not be overridden.

My reading of Tim's explanation seemed to find semantic benefit (and additional 
validity testing) within the Aries project. Marking the methods synthetic would 
seem to be sufficient (by our own proxy rules) for indicating that external 
proxying technology should keep their mitts off the methods. So, I'm not sure 
what additional benefit it's providing to external projects... If synthetic is 
sufficient, why over-specify? A minor (according to JIT developers) performance 
improvement? I'd welcome any hard performance data, BTW...

We've found one project that this *will* cause problems for (now fixed in their 
dev trunk). Will not be surprised as other problematic projects are found. For 
what benefit? 

Anyway, I'm done with the subject. Assume that David will be reverting his 
change

--kevan

Reply via email to