feat: fee foresight support

[LHerskind]

Fixes #10261.

Allow passing a timestamp for foresight. Also updates the getCurrentBaseFees to really return the fees that will be used at the next block. Essentially, what you currently see as needed to be included. I think keeping the name as is, is simpler to understand, as it is what you want to use if you are sending the tx "currently".

To see tests related to this, we have updated test_FeeModelEquivalence() such that it is predicting the fee up front and checking that it match (showcase foresight). And a separate test test__FeeModelPrune which will compute the fees with and without a future prune, showing that the foresight also considers the pruning.

[LHerskind]

• feat: fee foresight support #10262 👈 (View in Graphite)
• master

This stack of pull requests is managed by Graphite. Learn more about stacking.

[spalladino]

Suggested change

	const timestamp = BigInt((await this.publicClient.getBlock()).timestamp + BigInt(this.ethereumSlotDuration));
	const timestamp = BigInt((await this.publicClient.getBlock()).timestamp + BigInt(this.aztecSlotDuration));

Shouldn't we be moving to the next Aztec slot, as opposed to the next Ethereum one?

[LHerskind]

If using the aztecSlotDuration then if you are "in" a slot that don't have a block, you will be using the timestamp of the next slot, even if the next block could be within this slot.

Neither of them is really "perfect" for this, it should probably look at the next possible block, e.g., could be the current slot or the next if the current is already filled etc.

[LHerskind]

Made an update I think should handle it better. Now fetching the last block, and ensuring that we are at least at the next block from that. Ensuring that we won't get the fee for the already published block, but for the next possible. Only needed for the end-user, as some of these checks are already done earlier in the run from sequencing.

[github-actions]

Changes to circuit sizes

Generated at commit: 272fa983efdaed9e7b7d82e20415fff2718f5387, compared to commit: 9440c1cf3834eea380014d55eef6e81cff8ffee8

🧾 Summary (100% most significant diffs)

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
private_kernel_reset_4_4_4_4_4_4_4_4_1	+4,719 ❌	+14.57%	+19,236 ❌	+22.24%
private_kernel_reset	+5,633 ❌	+6.70%	+19,440 ❌	+3.15%
rollup_block_root	-168 ✅	-3.74%	-144 ✅	-0.01%
parity_root	-252 ✅	-5.01%	-216 ✅	-0.01%
rollup_merge	-168 ✅	-4.91%	-144 ✅	-0.01%
rollup_block_merge	-465 ✅	-3.87%	-451 ✅	-0.02%
rollup_root	-465 ✅	-3.88%	-451 ✅	-0.02%
private_kernel_tail	-6 ✅	-0.15%	-6 ✅	-0.05%
parity_base	-252 ✅	-5.86%	-216 ✅	-0.70%
rollup_base_private	-28,501 ✅	-14.42%	-41,353 ✅	-1.56%
private_kernel_inner	-785 ✅	-2.09%	-1,034 ✅	-1.82%
private_kernel_init	-741 ✅	-3.46%	-986 ✅	-2.86%
rollup_base_public	-91,453 ✅	-20.45%	-129,514 ✅	-3.24%
private_kernel_tail_to_public	-2,133 ✅	-12.27%	-3,044 ✅	-10.32%

---

Full diff report 👇

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
private_kernel_reset_4_4_4_4_4_4_4_4_1	37,106 (+4,719)	+14.57%	105,730 (+19,236)	+22.24%
private_kernel_reset	89,662 (+5,633)	+6.70%	637,479 (+19,440)	+3.15%
rollup_block_root	4,322 (-168)	-3.74%	2,739,521 (-144)	-0.01%
parity_root	4,782 (-252)	-5.01%	3,636,536 (-216)	-0.01%
rollup_merge	3,252 (-168)	-4.91%	1,826,915 (-144)	-0.01%
rollup_block_merge	11,544 (-465)	-3.87%	1,858,073 (-451)	-0.02%
rollup_root	11,528 (-465)	-3.88%	1,858,059 (-451)	-0.02%
private_kernel_tail	4,127 (-6)	-0.15%	12,509 (-6)	-0.05%
parity_base	4,049 (-252)	-5.86%	30,482 (-216)	-0.70%
rollup_base_private	169,174 (-28,501)	-14.42%	2,617,309 (-41,353)	-1.56%
private_kernel_inner	36,767 (-785)	-2.09%	55,848 (-1,034)	-1.82%
private_kernel_init	20,668 (-741)	-3.46%	33,520 (-986)	-2.86%
rollup_base_public	355,674 (-91,453)	-20.45%	3,867,937 (-129,514)	-3.24%
private_kernel_tail_to_public	15,253 (-2,133)	-12.27%	26,461 (-3,044)	-10.32%

[LHerskind]

Updated the artifact hash since it had changed after the noir sync