On Sat, Mar 7, 2009 at 7:57 AM, Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> wrote: > "Stephan Assmus" <superstippi@xxxxxx> wrote: >> I remember talking briefly about the subject with Ingo. Basically, >> IPv6 is not yet >> available by all providers. So the IPv6 code paths must be used >> optionally in >> Transmission. Otherwise the client would not work on most setups, >> right? >> So why is IPv6 being made a requirement at compile time? > > That basically just means that the OS must provide the headers at > compile time; since the client cannot use IPv6 yet, we do not have to > provide actual functionality. > I don't think this would be much work, and we need those headers anyway > at some point (and eventually a few function stubs). > It isn't a requirement at compile time, but rather throughout the codebase. The source files themselves are assuming IPv6 support. Here is the current file that fails to build : http://trac.transmissionbt.com/browser/trunk/libtransmission/net.c Here are some other files that mention "INET6": http://trac.transmissionbt.com/browser/trunk/libtransmission/blocklist.c http://trac.transmissionbt.com/browser/trunk/libtransmission/fdlimit.c http://trac.transmissionbt.com/browser/trunk/libtransmission/net.h http://trac.transmissionbt.com/browser/trunk/libtransmission/peer-mgr.c http://trac.transmissionbt.com/browser/trunk/libtransmission/peer-msgs.c http://trac.transmissionbt.com/browser/trunk/libtransmission/resume.c