On Tue, Feb 05, 2008, Amos Jeffries wrote: > You begin to see why I threw out all the initial attempts at IPv6 C-styles.
Heh. > Xuan and Rafael spent 4 months on trying squid-3 C-style and never got far > beyond the step-1 you are suggesting starting at. Well, if Step-1 is done and the code is better structured for it, then lets roll it into Squid-2.HEAD. We don't need to include a massive changeset that implements "IPv6" if its such a monumental task. > You don't mention the SNMP extension issues, external helper issues, > transparency and v6-port TOC issues, bridging issues, pooling issues > (which 3 still has!), FTP extensions, v4-mapped-v6 issues in v4-only > components. There are a LOT of things IPv6 needs to match IPv4 support in > squid. Oh yes, I know. But not a lot of those are required for website acceleration and thats probably a good place to start.:w > You could poach many of them from -3 I suppose, but to back-port it to > sockaddr castings would be more work than it took me to write in the first > place. Well, the way it looks at the moment, it'd be better to do somewhat what you've done - create a new opaque type and have a set of functions which replace the functionality they currently offer. Adrian
