Whats up with bitcoin core wallet locktimes

[arminsabouri]

This is an open question and reminder to my self to make a ticket upstream.
Following up on the discussions we had in #1445 .
It seems like nlocktime is set to the current height (anti-fee sniping behavior + 10% being backdated 100 blocks for some reason) when using send but defaults to 0 on other tx creating commands like createrawtransaction or walletcreatefundedpsbt. Seems like this behavior should be uniform.

[nothingmuch]

10% being backdated 100 blocks for some reason

iirc it's a random number up to 100 blocks in some of these cases, to introduce ambiguity (implying a tx was signed earlier than its first broadcast, which reduces temporal fingerprinting reliability somewhat)

Seems like this behavior should be uniform.

makes sense to me

[benalleng]

Opened an issue in core bitcoin/bitcoin#34987

[benalleng]

A question for us is whether we should add this backdating spread or will it just happen naturally with the nature of state updates in pj? It seems plausible that 10% of payjoins utilize the async nature to the fullest and are broadcast well after the initial psbt creation.

I guess if this is changed upstream we won't need to have any logic at all if the default is uniform in core. Then they can fight it out there whether the backdating is good or not.