While I agree with you n the problem, the solution is not really solving anything because it adds a condition to every single codepath after the fact. You are effectively adding 3 codepath while removing 2 here which is not a very clear win.

I removed the edit from the bot as it was not appropriate - pointing to the whole PR when a perfectly good change was available.

Split out the miner change and we can land the rest. Let's not delay due to open discussion.

Does this make diffs automatically?

I do not think this was a bad move. You'll need a TxMetrics and a BlockMetrics type and both will use ScriptExecutionMetrics in some way, but not the rest of interpreter.h .

I think this is okay because the sigops part is likely to change quite a bit soon, however, please make sure people can find the moved code when they backport.

The use of STANDARD_SCRIPT_VERIFY_FLAGS is a problem, everything else is all good.

If that was a bug, a test would really be beneficial.

rebase

Address comments

It would reverse the metric setting logic as to avoid a double negation. Everything else is good to go.

Update as per comments

The general approach look good. Obviously, Without changing ScriptExecutionMetrics, it's hard to ensure your invariants hold, so you may want to unpod that guy as well.

A few minor improvement, but it's getting there. You should consider starting to work on activation logic and tests.

Requesting change, but most of it are actually questions. The only thing I'm really suspicious of is the block template thing.

Requesting changes, but it's more like questions.

From what I gather, it is fairly obvious that you actually don't want a but are trying to make things work as a pod. The problem when you distribute responsibilities is that you always end up with something inconsistent down the road or a huge pile of spaghetti because various pieces of code needs to be kept in sync.

Rerun build