This sentence is quite unclear to me: which checks are you talking about? I don't think this new sentence is necessary at all to be honest, the requirements above already apply to all commitment types, and this paragraph only says that without `option_anchors_zero_fee_htlc_tx` you should be extra cautious when the feerate changes.

i think this might be formatted wrong? Here is what it looks like to me, i think the FAIL should be be indented under this? Sorry if I'm totally misunderstanding <img width="1062" alt="Screen Shot 2021-10-21 at 3 55 29 PM" src="https://user-images.githubusercontent.com/3514957/138355505-5ddd2b85-a853-4a00-be34-d07b46c8dc5f.png">

I don't think the spec needs to be this prescriptive. This is what we do in LDK, but others can/should do other things. I think we the spec should just talk generally about `feerate_per_kw` and note that in calculating you SHOULD provide some buffer for future feerate increases.

```suggestion - SHOULD fail the channel ```

```suggestion - SHOULD fail the channel ```

imo this whole section should be moved into the `update_add_htlc` requirements, as they're actions that are triggered by the receipt of this message.

This should be ```suggestion - if the `dust_balance_on_holder_tx` at the new `feerate_per_kw` is superior to the `max_dust_htlc_exposure_msat`: ``` The real question is if the new feerate pushes the balance over the exposure limit, not if the new feerate + buffer does. Pushing a feerate update where the new buffer ceiling is above the current should just fail to add any new dusty htlcs, but will leave the existing ones without a problem. Otherwise what was the point of having the buffer in the first place?