SCORU: carbonate!

We need to check that the SCORU code is correctly carbonated, and fix it otherwise.

Meeting notes here.

Good entry-points for this are the SCORU operations (defined in Operation_repr) used in Apply, because every SCORU code is in the scope of applying at least one of these operations.

• Benchmark when needed:
  • in snoop benchmarking closures, we should avoid definitions of the form let closure () = ignore p where p is of type _ Lwt.t because there is no strong guarantees about the actual execution of the promise (even though in practice this should work most of the time);
  • snoop does not take care of storage-related costs: hence a function that uses the storage can be benchmarked by snoop but we must manually add the cost of the storage accesses if there are not already taken care of by a carbonated storage.
• Add regression traces that will show gas changes.

Sub-issues have been created to tackle each operation separately:

• Sc_rollup_originate: #3291 (closed).
• Sc_rollup_add_messages: #3292 (closed).
  • Analysis previously performed:
    • Some carbonation may be left to implement (or we need to justify why not).
    • Needs reviewing.
  • But a new implementation is on the way (by Sylvain). We'll see then if there is some carbonation work left.
• Sc_rollup_cement: #3293 (closed).
• Sc_rollup_publish: #3294 (closed).
  • Analysis previously performed:
    • Some carbonation to add (Token operations).
    • Reviewed.
  • New strategy: macro benchmark instead.
• Sc_rollup_refute: #2926 (closed).
  • New approach: top-level over-approximation
  • =Carbonation in progress: !6241 (closed)= ((Replaced by smaller MRs).
  • Refutation game dissection !6937 (merged)
  • =Refutation game find tick !6945 (closed)=
  • =Refutation game start/stop !6966 (closed)=
  • =Refutation game proof (incomplete) !6967 (closed)=
• Sc_rollup_timeout: #3295 (closed).
• Sc_rollup_execute_outbox_message: #3296 (closed).
  • Consume gas for proof verification - 1 !7045 (merged)
  • Consume gas for proof verification - 2 (benchmark model) !5747 (merged)
• Sc_rollup_recover_bond: #3297 (closed).
• Sc_rollup_dal_slot_subscribe: #3298 (closed).
  • DAL team's responsability.
• And then, there is the issue about benchmarking the cost constants introduced when carbonating the above operations: #3976 (closed).