On Wed, Apr 19, 2023 at 12:53:56PM +0200, Claudio Jeker wrote: > With this diffs all the bits I committed before are connected and allow > for flowspec to be announced to other systems. > > I first thought I will put the flowspec_valid() inside flowspec_add() but > I changed my mind now and moved it into the imsg handler.
Whichever way you prefer. > flowspec_add() and flowspec_delete() are built a bit like > network_add/_delete but use their own special prefix functions. This is > because I did not want to implement all of Adj-RIB-In and Loc-RIB handling > just yet. > > All flowspec prefixes are stored in a static RIB tree (flowrib) and this > is used to push them into Adj-RIB-Out. This is for sure a hack but is > right now the smallest way to move forward without rebuilding a lot of > code on the input side. Seems fine for now. > The introduction of flowrib requries a change in one of the regress tests. > I have a diff for that ready. Go ahead, can't spot anything to complain about. ok tb
