Hi Eric, 

On 6/19/17, 5:44 PM, "BIER on behalf of Eric C Rosen"
<[email protected] on behalf of [email protected]> wrote:

>> It is somewhat strange to make protocol drafts standards track while
>> the architecture and encapsulations are experimental.
>
>I believe the documents are all supposed to be Experimental. However,
>this is not necessarily reflected correctly on the first page of each
>draft.

They seem to be all “Standards Track” right now.
>
>> The second example in the MPLS encapsulation draft implies a single
>> contiguous range of labels when, in fact, it must be encoded as 4
>> separate label ranges in the OSPF draft.
>
>That implication is unintended, as there is no suggestion that the
>labels L1, ..., L12 are successive numeric values.  At least, there was
>no intent to suggest that.  I can add a sentence to make that clearer.

Yes - that would definitely satisfy my comment.

>
>> I think It would be good then to tie the IGP encodings to the two
>> examples.
>
>I think it is appropriate in that example to point out that twelve
>labels must be allocated, each corresponding to a different <SD, BDL,
>SI> triple.  How those twelve labels get signaled is outside the scope
>of the encapsulation spec, and is totally a matter for the signaling
>documents to address.

I agree. After our protracted discussion, I did see that the OSPF draft
explicitly specifies "A unique label range is allocated for each BitStream
length and Sub-domain-ID.” However, I still think it would be clearer if
the protocol draft said something like: "For example, to advertise the
labels of the BitStrings in the second example in […], four labels ranges
would be advertised, one for each unique Sub-Domain and Bit String Length
tuple.”

Thanks,
Acee  

>
>_______________________________________________
>BIER mailing list
>[email protected]
>https://www.ietf.org/mailman/listinfo/bier

_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to