Re: SLAAC renum -- revised algorithm (Fwd: New Version Notification for draft-gont-6man-slaac-renum-04.txt)

[Fernando Gont <[e-mail]>]

On 12/3/20 12:14, Philip Homburg wrote:
>>> Please write drafts that actually work when implemented.
>>
>> Please be constructive.
> 
> You keep creating drafts that have issues and then when I point out
> those issues there is constant in your draft that can be increased and
> that makes the performance much worse.

You point out corner cases where things might go wrong. All protocols 
have that, starting from the fact that we normally employ best-effort 
transimission technologies, and hence there are no reliability warrants. 
And what I've mentioned is that even in such scenarios, one can be even 
more conservative if one wanted. -- For instance, I don't mind 
LTA_INVALID being 1800 sends, if that's the show stopper. Or even double 
that.

Performance does not get any worse. If anything, it takes longer to 
*garbage-collect* stale addresses. The high order bit is that you want 
new communication instances to work asap. And this draft achieves that.

Your input is indeed heard. We did add a whole section in response to 
your feedback.

Thanks,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492