This is super interesting. Thanks for such a detailed report and investigation!
I agree on the part that this is almost a “free parallelization enabler”. Which ofc comes in handy for proposals like parallel execution (which was also explored in Execution Dependencies).
I have some questions about the study and proposal.
- The EIP doesn’t really specify well why 2^24. Could this be more? or less? Also, why specifically this value and not another? Does it come from data? Or data comes from the number? Have you considered other values and if so, why?
- Do you see this as a Glamsterdam proposal? Seems low-cost/effort and with huge upsides and minimal affected contracts.
- Do you think it makes sense to expand the study to the last year or until pre-merge? Do you expect things will change much? (I’d say no, but curious on your take on why you studied the specific period you did).
- As per DOS-prevention, this would only prevent ammortizing the 21k gas of the tx across lots of actions. But aside from that, what’s the DOS prevention here? Users (without ammortizing the 21k gas) can still submit up to gas_limit consumption split in different txs. I’m not sure I fully understand the DOS point. Specially since users pay for gas anyways.