Aleksandr Bezobchuk
Aleksandr Bezobchuk
> Maybe we should also have some defensive mechanism that any kind of panic in grpc server shouldn't crash the node. I totally agree and I think like you said,...
I'm not sure I understand @amaurym. AFAIK, CW would not make any historical queries, but rather assumes the latest/current state. I like @aaronc's proposal of a Proto annotation. This way...
I've never seen or used Rapid, but the approach seems reasonable to me.
@amaurym I'm a bit lost as to how your points relate to non-determinism re queries: > server-side logic change (e.g. additional server-side validation of incoming request on the query handler)...
> Then our existing deterministic tests won't fail. Even though this is a state-machine breaking change. We would need to write new tests. Yeah, makes sense. I suppose there isn't...
@amaurym do you have any insight here?
Actually, I still think we want this @jackzampolin, but we should be wary of implementation. i.e support up to `n` transitive redelegations, where `n` is relatively small (e.g. 3-5).
Correct @rigelrozanski
Yes, but this is pretty low-pri, given everything else we have.
We should do this along with ensuring globally unique aliases.