Re: 3.0 timeline

On 20/11/12 16:05, Arran Cudbard-Bell wrote:

Eh, some of the proxy stuff is a bit explody on server down.

This has always been a pain to reproduce, too. Invariably when I got a proxy crash (on the old code) I would try to re-pro the problem by selectively making an upstream server fail, but it wouldn't crash again. Seems it was timing/state related :o( Hopefully Alan's rewrite will have quashed it; TBH part of the reason for wanting to run a pre-release is to show any bugs. I can arrange for one of our ORPS to run 3.0-pre and the other 2.2, which prevents shooting myself in the foot, but should hopefully get a bit of torture testing.