log fees in FillLog

[Lou-Kamades]

No description provided.

[Lou-Kamades]

this is match_base_lots per

openbook-v2/programs/openbook-v2/src/state/orderbook/book.rs

Lines 228 to 243 in 36eb9c4

	let fill = FillEvent::new(
	side,
	maker_out,
	best_opposing.node.owner_slot,
	now_ts,
	event_heap.header.seq_num,
	best_opposing.node.owner,
	best_opposing.node.client_order_id,
	best_opposing.node.timestamp,
	*owner,
	order.client_order_id,
	best_opposing_price,
	best_opposing.node.peg_limit,
	match_base_lots,
	);

[skrrb]

Please note that this amount will not correctly reflect the fees paid by the taker as it can be (slightly) greater than the actual fees paid: taker fees are computed from the total taken amount, not from the sum of the fill amounts, that is market.taker_fee_ceil(total_base_taken_wo_self_trade) <= \sum_i market.taker_fee_ceil(fill_amount_wo_self_trade_i)

[Lou-Kamades]

I don't fully understand this. Wouldn't the difference between incrementally calculating the taker fee vs. calculating using the total taken amount just be attributable to the rounding in market.taker_fees_ceil()?

I'd expect that total_quote_lots_taken and then total_quote_taken_native_wo_self would be equal to the sum over all fill events of the quote_native calculated in the fill event?

openbook-v2/programs/openbook-v2/src/state/open_orders_account.rs

Line 127 in 36eb9c4

let quote_native = (fill.quantity * fill.price * market.quote_lot_size) as u64;

If so, I think the rounding error is acceptable. Otherwise I'm afraid this change would be much more invasive since the taker fee can only be calculated after multiple fill events are emitted.

[skrrb]

exactly, the differences will be only attributable to the rounding in market.taker_fees_ceil(). just wanted to make it clear because I wasn't sure how much precision you would need for the events; maybe i'd suggests to add the suffix _ceil to taker_fess or something similar to make it completely explicit in the code.

On the other hand, total_quote_lots_taken will be equal to the sum over all fill events quote_native, so you could approximate the self-trade amount with the difference between the actual fees paid & the hypothetical fee that would be paid if no self-trade amount was discounted.

[Lou-Kamades]

Thanks for the explanation!