On 2020/03/04 00:55, Robert Scheck wrote: > On Mon, 02 Mar 2020, Theo de Raadt wrote: > > Can't we assume bird1 will go away eventually? > > Yes, sooner or later. Likely later for OpenBSD when looking to net/bird/1 > and net/bird/2 co-existence, thus ongoing BIRD 1.x compatibility might be > more important for OpenBSD than for rpki-client downstreams such as Fedora > (which only ships BIRD 2.x). And I expect I[SX]Ps to be quite conservative, > too. > > Could maybe somebody with real-world insights comment on this, please?
Definitely some IXPs run BIRD 2 already. IXP-Manager has supported 2.x in releases for ~10 months now and IIUC only does RPKI with BIRD 2.x. At least LONAP and INEX are running this. arouteserver only added BIRD 2 support recently so there's probably more of a mixture of versions among IXPs using arouteserver. As far as the OpenBSD BIRD ports go, I wanted to add 2.x early and at that point it was too new to fully replace 1.x so I just added it in parallel. But I don't think there is any particular need for rpki-client to support both. > > Or why don't those people convert json to their required format -- make > > it their problem? > > Are you referring to BIRD 1.x or 2.x here? I would prefer native BIRD 2.x > compatibility rather having to convert json around. Nevertheless, I'm not > sure if it's fair to remove the BIRD 1.x compatibility (which is currently > broken anyway). .. > Honestly, my downstream needs would be satisfied with "-B" leading to BIRD > 2.x compatible output only (avoiding new options and restructuring), but I > don't want to behave like a bull in a china shop. Thus in the end it's your > decision as upstream... If rpki-client BIRD 1 output is broken, IMHO it makes more sense to do just that i.e. replace broken 1.x output with working 2.x.