On Sun, Jan 02, 2011 at 07:06:15AM -0500, Jesse Vincent wrote: > On Sun, Jan 02, 2011 at 10:02:34AM +0000, Nicholas Clark wrote: > For me, it's not nearly as much about space savings as about > modularization. Being able to take, say, all of our socket > implementation into a possibly-replacable module makes it > much easier to, say, drop in an IPv6 compatible socket > implementation and removes another chunk of complexity > and code from the default runtime. I don't know where this myth is coming from that our socket implementation is not IPv6 compatible. It *is*. The core socket code is protocol agnostic, and Socket has had the AF_INET6 and PF_INET6 constants available for years. What has been missing are the IPv6 address resolution functions (being worked on) and the pack/unpack address functions (added before 5.12) But it's omission, rather than incompatibility. The Socket6 module on CPAN never needed to override/replace any of the functions in core. The "issue" was always that the core distribution alone was not enough to use IPv6. (And a legitimate issue that I'm glad we now have knowledgeable volunteers to fix) Nicholas ClarkThread Previous | Thread Next