<div style="font-family:arial,helvetica,sans-serif;font-size:10pt"><div style="font-family:arial,helvetica,sans-serif;font-size:10pt"><div style="font-family:arial,helvetica,sans-serif;font-size:10pt">On Wed, Nov 21, 2012 at 10:28 AM, Doug Barton <span dir="ltr"><<a href="mailto:dougb@dougbarton.us" target="_blank">dougb@dougbarton.us</a>></span> wrote:<br>
<div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div>I tend to agree with the spirit of this implementation (and for the record, I would have done HE differently in any case), but my immediate concern is for the "cached known-good IPv6" state when there is an intermittent problem on the link.</div>
</blockquote><div><br></div><div>Look at it this way: in IPv4, if you have intermittent problems on a link... well, it intermittently doesn't work. For thirty years, the answer to that problem has been, "don't give a host an address (or router, or DNS server, etc.) that doesn't work".</div>
<div><br></div><div>You could argue this is the way it should be. I suspect that app developers don't want to be in a world where every address may or may not work and they have to constantly probe destinations using all the host's addresses every time they connect. They would probably perceive that to be a waste of development effort and CPU time.</div>
</div></div>
</div>
</div>