Dynamic graphs are "possible" with metaphor, but there's more discussion to be had here.
One way to make "dynamic graphs" is to store small static graphs as a lookup. This is a hypothetical workaround but I don't have an empirical example of this.
Another is to audit individual node destruction and decide what needs to happen to the dependency graph. This makes an optimizing graph's job harder because destroying a node that is depended on downstream corrupts the execution sequence.
Dynamic graphs are "possible" with metaphor, but there's more discussion to be had here.
One way to make "dynamic graphs" is to store small static graphs as a lookup. This is a hypothetical workaround but I don't have an empirical example of this.
Another is to audit individual node destruction and decide what needs to happen to the dependency graph. This makes an optimizing graph's job harder because destroying a node that is depended on downstream corrupts the execution sequence.