- // In theory, we would be safe against unintentional channel-closure, if we only required a margin of LATENCY_GRACE_PERIOD_BLOCKS.
- // But, to be safe against policy reception, we use a longer delay.
- if (*outgoing_cltv_value) as u64 <= (cur_height + HTLC_FAIL_BACK_BUFFER) as u64 {
+ // If the HTLC expires ~now, don't bother trying to forward it to our
+ // counterparty. They should fail it anyway, but we don't want to bother with
+ // the round-trips or risk them deciding they definitely want the HTLC and
+ // force-closing to ensure they get it if we're offline.
+ // We previously had a much more aggressive check here which tried to ensure
+ // our counterparty receives an HTLC which has *our* risk threshold met on it,
+ // but there is no need to do that, and since we're a bit conservative with our
+ // risk threshold it just results in failing to forward payments.
+ if (*outgoing_cltv_value) as u64 <= (cur_height + LATENCY_GRACE_PERIOD_BLOCKS) as u64 {