Re: WGLC for BFD over Link Aggregate Group Interfaces - ends November 8

[Marc Binderberger <[e-mail]>]

Hello Mahesh,

> What is wrong with being more clear?

what you propose isn't more clear. And no, there is no specific meaning 
for "interface information" outside a specific implementation. I.e. 
it's implementation specific, which is why we are not trying to say 
more - we would get it wrong for the N+1 implementation (we hardly can 
know them all :-)

As I said, nothing wrong asking question and the answer is it is a 
generic "interface information" tat identifies the particular 
interface. To say more is already wrong for the specification. Even if 
your system uses ifindex (or whatever name an interface identifier 
number has in your implementation) then, as you mentioned, it doesn't 
mean the APIs provide this information. Which means an implementation 
for this draft knows the "interface" in a different way. And again I 
cannot specify what "different way" means, as this would be just one 
example I have in my mind.

The wording has been chosen very carefully ... .


Regards, Marc



On Thu, 31 Oct 2013 23:10:10 -0700, Mahesh Jethanandani wrote:
> Marc,
> 
> I do not know about you, but when I read interface there are specific 
> implications in my mind of what it means. To a certain extent I can 
> see that is in this response from Manav. It means something very 
> specific.
> 
> On Oct 27, 2013, at 4:37 AM, Bhatia, Manav (Manav) wrote:
> 
>> Can you explain the scenario where you think its not possible for a 
>> system to know the ifindex of the IP interface over which an 
>> incoming packet arrived?
> 
> My point is that if you believe that by "interface information" you 
> do not necessarily mean the interface (ifIndex) itself or that it is 
> a fairly generic reference to an identification of a member link of 
> LAG, then just say so. Or better still, say what it should mean. What 
> is wrong with being more clear? If I mis-read what was meant, so will 
> others.
> 
> Cheers.
> 
> Mahesh Jethanandani
> [e-mail]
> 
> 
>