optout

LGTM! Minor comments: - Maybe add a note to the `ComputeMerkleRoot` declaration, that the `hashes` input is expected to have one extra capacity (in case size is odd). Minor, as...

ACK c4293c7b21971e7ddd813ad3fbe574eb5605d60e With the recent commit restructuring the steps are clear. The code change is sound and looks well tested. I've left one non-blocking nit, no other comment.

> * It looks to me like the `expected_code_path` assert is confirming that the `std::logic_error` exception throws at least once and that test coverage is complete. I don't think it...

reACK 24d861da7894add47747eff69dd3fc71fbcdd7d0

Some ideas for RBF: - Split the current `Funded` channel phase into `FundingPending` and `Funded` - In `FundingPending` have the option to have not one channel, but several -- the...

RBF attempts differ only in transaction details, but I am also considering splicing, where pre-splice channel and post-splice channel -- with possibly multiple variants -- have to considered together. Pre-splice...

A current proposal for channel states/phases is described here: https://gist.github.com/optout21/2ead7a0f4e2c3d34aed1c9703c8cfb39

Note: Changing this behavior causes a single test case to fail: `ln::payment_tests::test_completed_payment_not_retryable_on_reload` (reproducible by not resetting `funding_transaction` in the original code, or keeping `funding_transaction_rebroadcast_flag` always false in the version of...

Alternative simpler solution in #3380 -- never clears the `funding_transaction` field, no new fields, which may result in additional rebroadcasts. One should be chosen -- I lean for the simpler...

This change request assumes that the whole `ChannelContext` struct is duplicated, which may not be the best solution. I put this on hold for now (back to draft, description updated).