On Mon, Nov 23, 2015 at 5:20 AM, Daniel Walton <[email protected]>
wrote:

>
> Just a heads up, we have a similar patch in the queue, it is part of the
> work we did on update-groups.  Will defer to Donald on when that will be
> posted for review.  I'm not saying one is better than the other but if
> Donald plans on posting it soon it might be worth holding off and taking a
> look at that and then we can figure out how best to move forward.
>

There is a rather key difference between this implementation and what we
have that is tied to update-groups. The table version really needs to be
per table - i.e., per AFI/SAFI and not a single value for the entire
instance. We have the table version in 'struct bgp_table' in
bgpd/bgp_table.h. The second aspect is that the version is also maintained
for each route node - thus, defined in 'struct bgp_node'. Of course, this
is not required when just keeping count of the overall change to a routing
table but is needed for the dynamic update-group implementation, to figure
out if two update-groups (subgroups) can be merged.

We also have the version as an 'u_int64_t' (unsigned long long) to handle a
monotonically increasing number.

I feel it may be worth holding off on this till the update-groups patches
are put up by Donald, especially because this is not directed towards
addressing any particular problem.



>
> Daniel
>
>
> _______________________________________________
> Quagga-dev mailing list
> [email protected]
> https://lists.quagga.net/mailman/listinfo/quagga-dev
>
_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev

Reply via email to