Closing fee_range uses invalid max value

[t-bast]

Hey,

I've been testing the quick close mechanism with the latest LDK, and there's one (small) incompatibility between eclair and rust-lightning. It's exactly similar to ElementsProject/lightning#4599 (comment). When rust-lightning doesn't care about the upper bound it sets max_fee_satoshis = 0xffffffffffffffff.

This is rejected by eclair, because we're strict on what values a satoshi field can hold. 0xffffffffffffffff is bigger than 21 million BTC so we consider it an invalid value. We could relax this but I'd like to avoid it if possible. That happens when rust-lightning is simply accepting a value inside eclair's proposed fee range, so I don't think it's necessary to set this to 0xffffffffffffffff, you could just return the same max_fee_satoshis you received.

WDYT?

[TheBlueMatt]

Sure, happy to change it to whatever, though ideally the spec would say that we have to keep it within a range if you're gonna check for it - much more sustainable than opening issues on all the implementations :)

[t-bast]

ideally the spec would say that we have to keep it within a range

I agree, there's no obvious place in the spec right now where we could require that "all bitcoin amounts sent in LN messages must be <= 21 million btc", but I'll add this before the next spec meeting.

[rustyrussell]

Set it to the channel size?

[TheBlueMatt]

Patch I hacked up just sets it to the remote balance, cause obviously the full channel balance isn’t spendable either :).

…

On Sep 8, 2021, at 19:33, Rusty Russell ***@***.***> wrote:  Set it to the channel size? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.