Re: [Gen-art] Review of draft-ietf-ccamp-flexible-grid-ospf-ext-07

[Pete Resnick <[e-mail]>]

On 6 Feb 2017, at 7:36, Daniele Ceccarelli wrote:

> Hi Jari, Pete,
>
> First of all thanks for the accurate review.
>
> All the nits/editorial comments are correct and will be fixed.
>
> Regarding the minor issue:
>
>> 4.1.1:
>>
>> The figure is a bit confusing: There might not exist a "Max Slot 
>> Width at Priority
>> 7" if bit 7 is clear in the Priority field, correct?
>> Perhaps it would be better to just show that as "..." or "Max Slot 
>> Width at
>> Priority n". The thing is, it might only be a single value, followed 
>> by the
>> padding, so having the second value in there might be misleading. 
>> (Perhaps
>> similar constructs are used in other MPLS docs and people will 
>> understand. But
>> it took me a while to figure it out.)
>>
>
> it was difficult to express in the figure the fact that only some of 
> the priorities are advertised. As you said it could be one, two or any 
> number up to 8.
> What about a single field (with ~ on the borders to indicate that it 
> can have variable length) saying "Max Slot Width at Prio n" and then 
> say that 16 bits are used for each prio and when an odd number of 
> priorities is used the field is padded to line up with multiples of 32 
> bits?

Yes, I think that would make more sense to me.

>> The discussion of Priority was very confusing for me. In the third 
>> sentence, do
>> you mean, "A bit is set (1) corresponding to each priority 
>> represented in the
>> sub-TLV, and clear (0) for each priority not represented in the 
>> sub-TLV"? I don't
>> understand the MUST/MUST NOT as you had it. And I don't understand 
>> the
>> last sentence at all. Are you trying to say, "The leftmost bit 
>> (priority level 0)
>> MUST be set, and priority level 0 MUST be advertised in the 
>> sub-TLV."?
>> Otherwise, I don't get it.
>
> It means that if the priority field is set to e.g. 10010010 in the 
> following you would find 4 fields indicating respectively: Max Slot 
> Width at Priority 0, Max Slot Width at Priority 3, Max Slot Width at 
> Priority 6 and since they are 3 a 16 bits padding to 32 bits.

OK, then I think my sentence is clearer for that: "A bit is set (1) 
corresponding to each priority represented in the sub-TLV, and clear (0) 
for each priority not represented in the sub-TLV."

> It also means that at least one priority must be advertised (i.e. 
> priority 00000000 is not allowed).

I get that part ("At least one priority level MUST be advertised"). It's 
the end I don't understand: "that, unless overridden by local policy, 
SHALL be at priority level 0." What does that mean?

>> I don't understand the MAY in the last sentence. Does that mean that 
>> I MAY
>> also set it to the highest possible nominal central frequency 
>> supported by the
>> link? I don't understand what that sentence is trying to tell me.
>>
>
> An example is provided in section 4.1.2, where the available range 
> goes from -2 to +8 but the range supported by the link goes from -9 to 
> +11. The sentence means that even if not available it could be 
> possible to indicate also n=-9 to indicate the starting point of the 
> range supported by the link.
>
>   " In this example, it is assumed that the lowest nominal central
>    frequency supported is n= -9 and the highest is n=11. Note they
>    cannot be used as a nominal central frequency for setting up a LSP,
>    but merely as the way to express the supported frequency range."
>
> I'm ok with dropping the sentence.

I think dropping the sentence would make the most sense.

> Thank you
> Daniele

Thanks for considering my suggested changes.

pr

>> -----Original Message-----
>> From: Jari Arkko [mailto:[e-mail]]
>> Sent: lunedì 6 febbraio 2017 12:32
>> To: Pete Resnick <[e-mail]>
>> Cc: [e-mail]; 
>> [e-mail];
>> [e-mail]; [e-mail]
>> Subject: Re: [Gen-art] Review of 
>> draft-ietf-ccamp-flexible-grid-ospf-ext-07
>>
>> Thanks for your review, Pete. Authors, any comments?
>>
>> Jari


-- 
Pete Resnick <http://www.qualcomm.com/~presnick/>
Qualcomm Technologies, Inc. - [phone]