Open glennhickey opened 5 years ago
I don't see this causing an issue on any existing implementations. In theory we could also accomplish this with ID 0, but it would require a bit of logic in implementations that don't explicitly encode the ID in the handle (namely XG). So yeah, I see the benefit.
Is there a reason this never got merged? I have an idea I want to try out, and I would need something along these lines for it.
An example of a null handle as proposed in #22. I think this is handy, especially when re-working existing code that may rely on checks of the form
Node* == nullptr
orid == 0
.It's a trivial addition to the interface, but imposes a condition that all implementations reserve the chosen value. Thoughts?