RE: Squid-2.4

> I think it is about time we settle on Squid-2.4 and get it
> released, but
> first I'd like a little wote to see if any of the pending
> changes should
> go in first:
>
> 1) aufs fixes to make it run
> 2) NTLM authentication
> 3) Transfer Encoding
> 4) SSL gatewaying
> 5) Sattelite link enhancements
> 6) Windows port using CYGWIN GNU toolchain
> 7) Authenticator protocol extensions
> 8) cbdata optimisations
>
> anything else I have missed?

I'd add

9) Have proxy_auth ACLs use splay trees rather wordlists

Rationale: the change is pretty small AFAICS, and should improve
performance quite a bit, especially in the case of enterprise-wide
caches like mine :-)

> From a political view it would be nice to have 2 also, but it contains
> some farily important changes in request processing and I am
> still a bit
> uncertain about of testing this code has received. (the part I am
> worried about is my part which is required by the NTLM changes, not
> actually the NTLM changes itself..)

Too little.
On the Squid side, Robert Collins is working on automatic helpers
start/kill, on the authenticator side, I have an uncommitted version
with a few nice features (automatic Domain Controller failover,
Domain Controller load balancing, saner command-line options), which
is to be thoroughly debugged.

> We do need to move along. 2.4 has been stomping by far to long now and
> is starting to get in the way for further developments.

I agree. I also propose to get 2.4 out NOW, and then concentrate
on NTLM for a very short 2.5 release cycle. 2.5 should only include fixes
and NTLM. For 2.6/2.7 we can then work on Digest authentication plus
other stuff.

> Yes, I do know that the 2.4 release is perhaps not in the shape one
> would like it to be for a release candidate, but my goal is
> now set for the next bakeoff.

"We have the technology. We can make it faster, stronger."
It's just a matter of time.

-- 
	/kinkie