IPv6 patch for BSD/KAME

Bryan O'Sullivan wrote:
It seems to be an oddity with OS X. Here's a well-timed patch that Thomas Schilling sent two days ago, which seems to be appropriate:
http://www.haskell.org/pipermail/libraries/2007-June/007644.html
That should indeed fix the build for BSD boxes. However, the underlying problem is that the existence of a s6_addr32 field in in6_addr isn't required by RFC 2553. So I'll cook up a patch that uses s6_addr, which *is* required, and hence ought to be more portable.

On 25 jun 2007, at 20.09, Bryan O'Sullivan wrote:
Bryan O'Sullivan wrote:
It seems to be an oddity with OS X. Here's a well-timed patch that Thomas Schilling sent two days ago, which seems to be appropriate: http://www.haskell.org/pipermail/libraries/2007-June/007644.html
That should indeed fix the build for BSD boxes. However, the underlying problem is that the existence of a s6_addr32 field in in6_addr isn't required by RFC 2553. So I'll cook up a patch that uses s6_addr, which *is* required, and hence ought to be more portable.
Yes, that would indeed be the cleanest solution, since that patch actually accesses internal data structures which may change add any time. I wonder, though, if it would hurt performance if accesses were changed byte-based. / Thomas

Thomas Schilling wrote:
Yes, that would indeed be the cleanest solution, since that patch actually accesses internal data structures which may change add any time.
I doubt they'll change any time :-) All of those APIs have been stable for a long time.
I wonder, though, if it would hurt performance if accesses were changed byte-based.
Very probably, but it's such a non-performance-related area of the code that I don't think you could measure the difference without a very special-case microbenchmark.
participants (2)
-
Bryan O'Sullivan
-
Thomas Schilling