Just use UAHF_START_TIME . If you have something that rely on specific time, your test is going to be flacky anyway.

I added this during initial construction of the abc-rpc.py changes.
Ended up not using it, but since it is a useful default I retained it for the Python framework.

As requested, I added a block to limit the LOCK

Agreed, removed

In principle I agree that it would be good to move such validation checks into SetUAHFStartTime() .

In practice, this comes with some complications - one needs to pass MTP into the validation method, but cannot do so from the call site at init.cpp:AppInitParameterInteraction() , since that is way too early. So that would require a workaround of introducing a minimum UAHF start time to pass for the non-existent MTP at that point, or have two separate validation methods entirely, one for RPC and one for command line parameter checking.

Last status on Slack discussion was to keep the validation in the RPC call for now.

So I am keeping them here for now.

Done - moved to after the excessive block methods

I fixed this call which was using the wrong offset (+ 2h 1s) instead of the +1s that the comment indicates.

The functions above were re-used from abc-p2p-fullblocktest.py as we discussed that factoring out a large-block generation into a common framework module could be done in a subsequent step.

generate_block() below is adapted from next_block() , with added control of the block's nTime so that I can activate the fork easily while also creating controlled block sizes at the same time.

I didn't modify this yet because this is more complex / tied in to test design which needs to test some steps before the fork is active, then activate it and do some checks afterwards (e.g. that UAHF start time can no longer be changed after the fork has happened).

Initially I didn't set the time on test node to shortly ahead of fork, and then I ran into various rejection cases where when I mined the "fork pre-activation" blocks, they would get rejected.

So I set the node up to be a few seconds ahead of the fork time now, so that I can mine the blocks and push the MTP to activate comfortably.

This is very stable as it is not using any machine time - the generated blocks also get their nTime set appropriately.

I could try and simplify this test though, but it's going to take me more experimentation.

This is for the genesis block case.

For the general case in l.145, if no ntime is passed, this is exactly what is done (use previous time +1 for the new block).

You only need to tweak one block to set it at the HF time and then pile up.

Yes, the test should be simplified in regard to how it uses this method. I'm looking into that.

You should check for the nothrow that the config has been updated accordingly.

Thanks, was bad naming.

Fixed

Done

Refactored

Refactored

Fixed, went over the file and did this plus replacing any assert() with Boost checks.