On Thu, 2019-11-21 at 15:37 +0800, Ben Schwartz wrote:
> I think we are talking about performance in a situation that (a) is so 
> pathological that it will almost never happen and (b) at worst, will only 
> slow down failure, not success.  For that reason, I would avoid introducing 
> any additional protocol complexity in pursuit of optimization.  I think our 
> simplest and most appealing option would be to treat EDE exactly like any 
> existing EDNS Option (i.e. set the TC bit).
> 
> To be clear, we are talking only about the case where a response "fits" until 
> the EDE is added, after which it exceeds the limit and is truncated.  If 
> optimizing this situation is important, I would suggest adding a requirement 
> to the EDE draft that EDE be the last option in OPT.  Then as a client, I can 
> easily detect this situation, because the truncation point is in the middle 
> of the EDE option.  The client logic then is very simple: if I don't care 
> about EDE, I can ignore the TC bit.

Please do not write specifications that require software to receive a corrupt 
packet (in this case, one where the RDLEN of the last record extends beyond the 
end of the packet) and draw any conclusion from it other than 'this is a 
corrupt packet'. Also, please do not truncate packets on arbitrary byte 
boundaries.

In other words, please always send well-formed packets that contain what they 
say they contain in terms of Answer/Additional/Auth counts, and the sizes of 
those individual records.

Kind regards,
-- 
Peter van Dijk
PowerDNS.COM BV - https://www.powerdns.com/
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to