Hi Donald, +1 We need this in our implementation of BGP-LS that used an AFI = 16388 for Traffic Engineering family. Dealing with the old implementation based on a table introduced to much size and CPU consumption in our case. Regards Olivier Le 05/12/2016 à 17:16, Donald Sharp a écrit :

David and I just had a quick discussion on AFI/SAFI handling in Quagga. Capturing result of conversation here:

1) Change #defines in zebra.h of SAFI_XXX to an enum as well as modifying them to SAFI_INT_XXX -> the _INT_ is to tell the world that these are internal values only.

2) lib/prefix.[ch] has code to map internal/external AFI/SAFI values.

I am going to submit patches for #1 and #2 hopefully in the next few days.

This mapping between internal and IETF AFI/SAFI values is there to allow bgp to not have a sparse array for peer information. I think that there is desire to move away from this solution. We do not have anything designed at this point beyond vague hand waving :)

donald

_______________________________________________ cmaster-next mailing list cmaster-next@lists.nox.tf https://lists.nox.tf/listinfo/cmaster-next

-- logo Orange <http://www.orange.com> Olivier Dugeon Senior research engineer in QoS and network control Open Source Referent Orange/IMT/OLN/WTC/IEE/OPEN fixe : +33 2 96 07 28 80 mobile : +33 6 82 90 37 85 olivier.dugeon@orange.com <mailto:olivier.dugeon@orange.com>