A few updates, as we’ve all been busy:
- Amos is pushing forward his IPv6 work slowly; its probably going to wait for Squid-3.0 to be released and made stable but its making progress
- I’ve done some profiling on Squid-3’s performance and I’m working on a few cheap methods to gain a few percent by not zero’ing buffers that don’t need zero’ing – another possible Squid-3.1 candidate
- We learnt how C++ will “upgrade” types where the correct constructor is available – part of some performance regressions in Squid-3 (and the more noticable crashes!) popped up after some changes were made to neaten the code up…
- Duane and Alex are still working hard to bring Squid-3’s behaviour close to stable as they can – hopefully a Squid-3.0 STABLE release will occur soon!
- I’ve been working a little on /dev/poll (Solaris, IRIX) support for Squid-2 and I think I’ve finally solved the behaviour issues. This should drastically improve Squid’s performance under Solaris and IRIX.