We can guess that the jumpdest analysis is not needed by inspecting the first opcode. This heuristic has some false negatives but is very cheap and covers EOF. The proof it is correct is done at compile-time.
Bigger idea behind this is to only use the heuristic not to waste time on jumpdest analysis and delay EOF header parsing until execution. This makes the code analysis EVM revision agnostic. Maybe someone can push this idea further.
We can guess that the jumpdest analysis is not needed by inspecting the first opcode. This heuristic has some false negatives but is very cheap and covers EOF. The proof it is correct is done at compile-time.
Bigger idea behind this is to only use the heuristic not to waste time on jumpdest analysis and delay EOF header parsing until execution. This makes the code analysis EVM revision agnostic. Maybe someone can push this idea further.