Re: [Idr] //Re: Status changes as of 6/21/2022

[Jeffrey Haas <[e-mail]>]

> On Jun 23, 2022, at 5:35 PM, Robert Raszuk <[e-mail]> wrote:
> 
> Hey Jeff,
> 
> > discusses how forwarding interactions stack together and start becoming tricky to reconcile.
> 
> Indeed ... and as we see here such interactions are already a possibility. One can only be scared of what is going to happen when we add v2 to the mix :) 

Ideally we figure out as part of v2 how to avoid some of the current headache.

As an example, imagine Wenying's proposal where the flowspec route contains a color, a redirect-to-ip, a redirect-to-vrf, and a rate limit.  What order do they get applied?  What about if there was an incompatible path-redirect in there?  Etc.


> - - - 
> 
> With that - while out of the topic - one sentence caught my attention: 
> 
> "The code points for Flowspec v1 and v2 are likely to be the same."
> 
> Not 100% clear on what you meant by "are likely" ..
> 
> If you mean reusing codes in v2 which are already defined in v1 then this has some pros and many cons. But perhaps you meant that v2 may just use currently defined codes (in v1) not that we will define them again in v2 with the same code points. 

My text was intended to mean the code points for existing flowspec v1 match criteria should be carried over into the initial set of flowspec v2.  It'll help keep people's code cleaner.

-- Jeff