> On Wed, Oct 27, 2010 at 5:26 PM, FRLinux <frli...@gmail.com> wrote: > > On Wed, Oct 27, 2010 at 9:45 PM, Theo de Raadt <dera...@cvs.openbsd.org> > > wrote: > >> The design process followed by the NFSv4 team members matches the > >> methodology taken by the IPV6 people. =A0(As in, once a mistake is made, > > > > Sorry, I'll bite. What exactly is wrong with IPv6 here? I gathered > > from this list not a lot of developers here like it, but I still don't > > get it. Please educate me (this should be enlightening). > > Instead of fixing the one problem with v4, they decided to fix a > thousand additional "problems" that nobody really cares about.
in that regard i disagree, but perhaps only in tone. With ipv6, they decided to create a bunch of new problems that people now find they care deeply about - they created a totally new problem by avoiding arp. the benefit of their layer-2 discovery mechanism has been absolutely zero; the best unit of measure for the cost of that decision is "decades". - they created a new problem by punting global routing to "further study" (in this, they showed that they had deep familiarity with appletalk and ipx). - they created an entirely new and huge problem (destroying SIOCGIFCONF backwards compat hurt IPV6 deployment in operating systems on a massive scale) by not making their sockaddr be a power of 2 in size. it sounds silly, but it turns out it is the kind of thing which matters. when they were told of this problem (very early on) they said something like "oh, but we already have 3 engineers in the world running their own ipv6 test code, so it is too late to change that". this is the specific mindset which results in layers of bad decisions papered over top of each other. shit which comes out of research organizations all tends to suck these days, doesn't it. or perhaps it always did (OSI networking, ipv6, same same). i have theorized in the past that the problem we face is that an insufficient number of axe murderers are attending those kinds of research meetings.