Re: [MERGE] sourcehash and userhash peering algorithms

On lör, 2008-07-12 at 15:01 +0200, Kinkie wrote:

> for what it's worth, +1.
> Maybe they could be abstracted more (need to check the patch better,
> will do once it's in), but it's a step forward already.

Sure it can be abstracted better, for example with a shared CARP-type
template above, with the implementations only providing the keying (1-4
lines each unique in carp.cc/peer_sourcehash.cc/peer_userhash.cc)

Have also thought about maybe only using a single instance in the peer
struct for the carp type data, but it's useful to combine at least
userhash with one of the other two as fallback (usually sourcehash)...

Regards
Henrik