Re: [Roll] terminology

["Pascal Thubert (pthubert)" <pthubert at cisco.com>]

Agreed. 

Rank is scoped within a sequence.

A node cannot use a passed parent (a parent that exposed a passed
sequence). 

A node may use a future parent (a parent with a future sequence) but it
has to appear as a leaf to that parent in that DODAG instance by setting
the rank to infinite. This is done to avoid a false positive detection
if the rank relationship changes with the sequence.

Cheers,

Pascal

>-----Original Message-----
>From: roll-bounces at ietf.org [mailto:roll-bounces at ietf.org] On Behalf Of
Julien Abeille (jabeille)
>Sent: mercredi 4 novembre 2009 17:36
>To: Jonathan Hui; Philip Levis
>Cc: roll ROLL
>Subject: Re: [Roll] terminology
>
>Hi all,
>
>I am inline with Philip. Thanks for the precisions by the way.
>Regarding sequence numbers, for me 2 nodes with 2 different seq numbers
>are in different DAGs, hence we do not need to compare them as far as
>height is concerned.
>
>Best,
>Julien
>
>> -----Original Message-----
>> From: Jonathan Hui [mailto:jhui at archrock.com]
>> Sent: mardi 3 novembre 2009 20:58
>> To: Philip Levis
>> Cc: Julien Abeille (jabeille); roll ROLL
>> Subject: Re: [Roll] terminology
>>
>>
>> On Nov 3, 2009, at 11:20 AM, Philip Levis wrote:
>>
>> > This seems like an excellent simplification to me.
>> >
>> > The one weird thing that occurs is that people often talk
>> about going
>> > "up" the tree as towards the root, which is the opposite of a real
>> > world tree. E.g., depth first search. So I'd suggest we say
>> >
>> > Up: towards the root, decreasing rank
>> > Down: towards the leaves, increasing rank Higher node: lower rank
>> > Lower node: higher rank
>> >
>> > I think it's important to define higher and lower in terms of Rank
>> > since Rank is what's actually used. Otherwise you get into a weird
>> > situation comparing disjoint subtrees.
>>
>>
>> If we want to be precise about the terms, then "sequence
>> number" needs to appear somewhere.  For example, A may be
>> "higher" than B because it has a newer sequence number -
>> which trumps any difference in rank.
>>
>> --
>> Jonathan Hui
>>
>>
>_______________________________________________
>Roll mailing list
>Roll at ietf.org
>https://www.ietf.org/mailman/listinfo/roll