If the backend analysis is able to serialize its entire state into a snapshot (this is more than just the abstract states it has computed, it is also things such as the worklist and the configuration options). Then many interesting features become possible.
[ ] timeline of snapshots to see how the analysis progresses
[ ] graph timeline of numeric values from the snapshots, e.g. worklist size, memory use
[ ] diff of two snapshot (i.e. "what has changed in the last 1000 node transfers)
[ ] rendering of the snapshots on a different machine (or just at a later timer)
[ ] restart of the analysis from a snapshot (see #2)
[ ] restart of the analysis from a snapshot - with changed configuration options (this is less related to the inspector)
If the backend analysis is able to serialize its entire state into a snapshot (this is more than just the abstract states it has computed, it is also things such as the worklist and the configuration options). Then many interesting features become possible.