> On Jul 22, 2025, at 8:13 AM, Tony Przygienda <[email protected]> wrote:
> 
> 1. yes, this needs work for the archeological ASCII support ;-) on out plate
> 2. fine, no issue with "non-prunner" and we can use N and N| for it. 
> 3. per Tony Li 
> 4. a CDS will be always computed, either in a distributed fashion or a 
> centralized fashion. If an algorithm doesn't compute CDS somehow how will it 
> know it doesn't partition the component, Acee?  And as 3., CDS is a well 
> known graph theory term, in fact I'm already loose in the language since it's 
> a edge connected dominating set   
> https://en.wikipedia.org/wiki/Connected_dominating_set

I know very well what a CDS and this is why I suggested "CPS" rather than 
"component". One could argue that the CPS is not a unidirected and hence not a 
"component" since the routers on either side of the edges the CPS sub-graph 
must support the same flooding algorithm. 

I still think CDS is orthogonal to CPS in this context since the CPS is a set 
of routers running the same flooding algorithm and CDS is the set of routers 
that are required to flood in a class of flooding algorithms. You could have a 
simpler flooding algorithm that don't require CDS computation but simply uses a 
heuristic to determine who floods and who waits to see if they get the LSP from 
their neighbor. Anyway, I guess it doesn't hurt that much to leave it in the 
document. 


> 
> --- 
> 
> 1. this is how you annotate things in graph theory. there may be multiple 
> components running A which is an algorithm and A| is the CDS algorithm A 
> builds. standard way is to say A|_1 and A|_2 really but since that looks 
> super ugly on ASCII I took the 2nd better choice which is A|' and A|''   . 
> Let's not be smarter than 300+ years of mathematicians that dealt with that 
> stuff.

Well, if you're looking at the .txt version that most people are reviewing in 
the meeting materials, it doesn't make any sense. With the figure, you can 
discern that A|' and A|'' are separate instances of a CPS running flooding 
algorithm A. However, I'm not sure such a simple concept benefits from the "A|" 
notation. Maybe you could at least add a legend to the figure to explain this 
to those reading it for the first time. 

Thanks,
Acee




> 2. sure 
> 
> 
> --- tony 
> 
> On Tue, Jul 22, 2025 at 12:28 PM Acee Lindem <[email protected]> wrote:
> Speaking as WG member:
> 
> It seems my comments on this draft continue to go without response. I would 
> have hoped that at least some of them would have been addressed during WG 
> adoption. 
> 
>    1. The draft is optimized towards _markdown_ and pdf rendering. It is 
> should be optimized towards text since that is what everyone is reviewing 
> when they download meeting materials. 
>    2. I've given up on not using the term "pruner" for "flooding algorithm" 
> as I can tell that there is "pining for pruning". However, for no flooding 
> algorithm, please do not use the term "zero pruner" and certainly not "zero". 
> Rather use, "no-pruner" or "non-pruner".  In fact, the term "zero" may not 
> even meet the IETF requirements for inclusive language. 
>    3. The invented term "connected component" is very confusing. We already 
> have CDS so why not "CPS"  for "Connected Pruner Set"? 
>    4. While we're talking about CDS, I think that section 2.1.3 is 
> orthogonal. There could be simple flooding reduction algorithms that do not 
> compute a CDS.
> 
> 
> Other minor comments:
> 
>     1. I'm not fond of the terminology of A|, A|', A|'', B|', etc. Is this 
> necessary? This could be just be flooding algos (i.e., pruners) A, B, and N 
> (for none). 
>     2. Should the "Contributors" section be "Acknowledgements"? 
> 
> 
> Thanks,
> Acee
>     _______________________________________________
> Lsr mailing list -- [email protected]
> To unsubscribe send an email to [email protected]

_______________________________________________
Lsr mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to