Tests use different price for refund, but the issues mentions incorrect gas price used by Stylus programs. In which cases will there be a difference ? Is it worth to add a test for that?

I've checked the test Stavros added, and it properly fails when opcodeGasPrice is not fixed.
The issue was identified by seeing the diff in passed EvmData to Stylus between Nitro and Nethermind https://github.com/NethermindEth/nethermind-arbitrum/blob/main/src/Nethermind.Arbitrum/Arbos/Programs/StylusPrograms.cs#L186

Theoretically, we can add test triggering this through Stylus contract, and this VM hook, and call to SDK method... gonna be fun

@damian-orzechowski I think the test I added adds sufficient test coverage for this and fails when opcodeGasPrice is not fixed.

This test verifies the dual gas price behavior

• EVM context (Stylus programs) sees originalGasPrice (1500 = base 1000 + tip 500) via the GASPRICE opcode
• Payment/refund uses baseFeePerGas only (1000) after tip is dropped

Do you think we should add more testing around it?

Minimum allowed line rate is 60%

@damian-orzechowski I think the test I added adds sufficient test coverage for this and fails when opcodeGasPrice is not fixed.

This test verifies the dual gas price behavior

• EVM context (Stylus programs) sees originalGasPrice (1500 = base 1000 + tip 500) via the GASPRICE opcode
• Payment/refund uses baseFeePerGas only (1000) after tip is dropped

Do you think we should add more testing around it?

Ok - let's merge then