Discussion topic for EIP-8053; PR
Abstract
This proposal introduces the milli-gas counter as the EVM’s internal gas accounting. Gas costs are defined in milli_gas and internal EVM gas accounting is entirely carried out in milli_gas. At the end of transaction execution,milli_gas is rounded up to gas. Gas limits and transaction fees are still computed and verified using the current gas_used counter. This new counter enables a more precise accounting of cheap compute without impacting UX.
it seems to me that the EIP as currently published conflates whether the GAS opcode ought to report the higher-level divide-by-1000-round-up gas concept or the lower-level milli-gas concept. as a consequence, the following statements are confusing:
At the same time, after the transaction is executed, evm.milli_gas_left is rounded up to evm.milli_gas_used = ceil(evm.milli_gas_used / 1000).
seems like this should state evm.milli_gas_used = ceil(evm.milli_gas_used / 1000) * 1000
The GAS opcode reports the available gas in milli-gas units.
*CALL opcodes receive the amount of gas to send to the sub context to execute in gas
latter two, when read literally, suggest that the common idiom of GAS => CALL results in forwarding 1000x (obviously not realizable, for several reasons) the gas available in the current context.
for myself, as a dApp programmer, I recommend keeping the semantics for *CALL as written in the text of the EIP, but modifying the semantics of GAS such that it reports floor(milli_gas_left / 1000), to preserve rough compatibility with existing dApps as much as possible. this hides the details of milli-gas implementation from the VM to a significant degree. with this modification, I can sincerely endorse this EIP as a Good Idea™
as a footnote, I also recommend that the EIP-150 all-but-one-64th rule be applied to gas and not to milli-gas before the amount of gas forwarded is converted to milli-gas inside the newly-created context. this seems like the obvious way to do it given the text of the EIP, but it seems to me that it ought to be explicitly stated for the avoidance of any doubt.