feat(lunarbase): support v0.4.1 pools and BSC indexing by ThomasAqu1nas · Pull Request #1445 · propeller-heads/tycho

ThomasAqu1nas · 2026-09-15T11:45:22Z

Pool v0.4 applies punishment to the current swap and changes the directional fees used by later quotes. The previous LunarBase integration used the older math and state schema, so updating the ABI alone could leave simulated quotes inconsistent with the pool.

This change updates LunarBase to Pool v0.4.1 and adds BSC indexing:

Use lunarbase-pmm-math 0.4.1, the full uint160 anchor, maximum punishment, and indexed caller fee policy.
Carry reserves and directional fees through consecutive simulated swaps; refresh oracle age only on operator updates and advance it with blocks.
Index the new events and bootstrap complete state from verified Base/BSC parent snapshots. Keep the full runtime ABI while generating bindings for the nine consumed events.
Add the BSC manifest, replay fixtures and the necessary test-harness registrations.

Substreams package 0.2.0 changes the state schema. It requires the matching simulator and a fresh replay for the configured quote caller.

Validation at 4e1b7a0bf4e0e18423bb32d999f1937f7a79da8b in the fork:

Rust CI: 1,169 unit tests and 141 PostgreSQL tests passed; formatting, strict Clippy, Rustdoc and no-default-features checks passed. The Retry-After timing test that failed locally passed here.
Substreams CI: LunarBase native/WASM builds, lint and all 19 tests passed.
Cargo machete and PR-title validation passed.
Locally, all 34 LunarBase simulation tests passed with all features. The current remote workflow filters do not execute these tests.
A trial merge with upstream d58c29d2ab71fb3821f4ce4eff22d3edf0803583 has no conflicts and retains only the 30 changed files.

Draft validation limits:

Security Audit fails on RUSTSEC-2026-0285 for rustls 0.23.31, inherited unchanged from the branch base. This commit changes only lunarbase-pmm-math in the root lockfile. Current upstream already uses rustls 0.23.45.
Full Substreams/RPC integration has not run. The fork has no archive RPC/Substreams secrets; its integration and EVM/RPC runs were cancelled. Production BSC executor registration is unverified.
The existing integration composite action exports PROTOCOL, while Docker Compose reads PROTOCOLS; this also needs to be reconciled for a complete range run. No general workflow changes are included here.

Update punishment math, indexed state, and caller-dependent fees while preserving oracle freshness across blocks and consecutive swaps. Add verified Base/BSC bootstrap snapshots and BSC test-harness wiring.

socket-security · 2026-09-15T11:46:12Z

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	cargo/lunarbase-pmm-math@0.2.4 ⏵ 0.4.1

View full report

zach030 · 2026-09-23T07:29:45Z

 ) -> tycho::ProtocolComponent {
    tycho::ProtocolComponent::new(&component_id(pool))
        .with_tokens(&[token_x, token_y])
+        .with_attributes(&[("quote_caller", quote_caller)])


What does quote_caller do? Can we just assume the Tycho Router fee is zero?

I'd suggest removing quote_caller in the substreams and quote_caller_whitelisted in the simulation. We can just assume the Tycho router is whitelisted. Otherwise, if we ever change the router, we'd have to update the substreams and do a full resync, which is a lot of extra work.

…g in bootstrap process

github-actions · 2026-10-05T11:47:00Z

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

ThomasAqu1nas · 2026-10-05T13:11:33Z

I have read the PropellerHeads CLA and I hereby sign the CLA

zach030 · 2026-10-06T06:46:30Z

+        Direction::XToY => (q96, anchor),
+        Direction::YToX => (anchor, q96),
+    };
+    let probe = (BigUint::from(reserve_out) * &numerator / &denominator) * numerator / denominator;


The empty-input-reserve probe can report zero limits for a tradable direction because these two divisions round down, while quote_limit only reduces the input after a rejected quote.

I reproduced this with a fresh, unpaused pool, zero fees and punishment, reserve_x = 0, reserve_y = 1, and anchor_price_x96 = 3 * Q96 / 4. For X -> Y:

This probe evaluates to floor(floor(1 / 0.75) / 0.75) = 1.

Quoting input 1 produces zero output, so quote_limit halves the input to zero and returns (0, 0).

However, input 3 is valid: the actual quote is floor(floor(3 * 0.75) * 0.75) = 1, and quote_exact_in(3) succeeds.

A temporary regression test confirmed that the direct quote succeeds while get_limits returns (0, 0). Consumers relying on these limits will skip this tradable direction.

Could we derive the probe from the quote's exact rounding rules, or distinguish an input that is too small to produce output from an input that exceeds the available liquidity, rather than halving both cases?

[P2] Reconfirmed on b68828334: this also affects the reverse direction and small nonzero input reserves.

A temporary regression test exercised fresh, unpaused states with zero fees and punishment, output reserve = 1, and these four cases:

Direction Anchor Input reserve quote_exact_in(3) output get_limits

X → Y 3 * Q96 / 4 0 1 (0, 0)

X → Y 3 * Q96 / 4 1 1 (0, 0)

Y → X floor(4 * Q96 / 3) 0 1 (0, 0)

Y → X floor(4 * Q96 / 3) 1 1 (0, 0)

With input reserve 1, the soft_limit branch starts at 2, which is still too small to produce output. Halving after rejection then incorrectly concludes that the direction has no capacity. Consumers using these limits will skip a tradable direction.

Please make the probe/search account for the minimum input needed to produce positive output, distinguishing undersized quotes from quotes exceeding available capacity. Regression coverage should include both directions and both zero and small nonzero input reserves. These are synthetic rounding-boundary cases, not observed mainnet failures; the existing 36 LunarBase tests pass but do not catch them.

…ltiplier - Removed `quote_caller` from bootstrap states and protocol components. - Eliminated `blacklist_fee_multiplier` and related events from the ABI and state management. - Updated JSON bootstrap files to reflect changes in attributes. - Adjusted integration tests to account for the removal of quote_caller. - Cleaned up event handling logic to ignore deprecated events. - Simplified configuration parsing by removing quote_caller requirement.

…nimum trade limits

zach030 · 2026-10-13T12:24:45Z

+  # This historical range predates router whitelisting. Execution comparison
+  # requires a whitelisted router; recapture a later range for on-chain parity.


[P2] Update the Base execution fixture to satisfy the whitelist assumption

The simulator now always uses fee multiplier 1, but this range still ends at block 51297765, where the configured Base router (0xAbA5B53b03eAfaD1C5fc8BD5Fc765fC85Bb3de67) is not whitelisted and pays multiplier 100. Execution comparison remains enabled, and the harness does not override LunarBase's whitelist. Adding this comment does not make the configured test environment satisfy the new assumption.

I read the pool state and quotes at block 51297765, using that router as from, and passed the same state into the current simulator:

Direction / input On-chain quote Current simulator

X → Y / 0.001 ETH 2,390,779 2,511,757

Y → X / 1 USDC 393,595,800,163,007 397,890,571,043,066

Both differences exceed the harness's 0.5% execution tolerance. Passing multiplier 100 into the math library reproduces the on-chain outputs exactly, isolating the caller-policy mismatch. These are block-pinned quote comparisons; I did not run the full range/fork execution test.

Please recapture the parent snapshot and range for an execution router verified to have multiplier-one policy, or explicitly configure and validate a whitelisted fork environment and document that state override. Keep the two-direction execution comparison enabled so this test validates the model it now assumes.

feat(lunarbase): support v0.4.1 pools and BSC indexing

4e1b7a0

Update punishment math, indexed state, and caller-dependent fees while preserving oracle freshness across blocks and consecutive swaps. Add verified Base/BSC bootstrap snapshots and BSC test-harness wiring.

github-project-automation Bot moved this to Todo in Tycho Sep 15, 2026

github-project-automation Bot added this to Tycho Sep 15, 2026

ThomasAqu1nas requested a deployment to integration-tests September 15, 2026 11:45 — with GitHub Actions Waiting

zach030 reviewed Sep 23, 2026

View reviewed changes

refactor: streamline parent hash validation and enhance error handlin…

b688283

…g in bootstrap process

ThomasAqu1nas requested a deployment to integration-tests October 5, 2026 11:47 — with GitHub Actions Waiting

zach030 marked this pull request as ready for review October 6, 2026 05:58

zach030 reviewed Oct 6, 2026

View reviewed changes

ThomasAqu1nas requested a deployment to integration-tests October 13, 2026 07:26 — with GitHub Actions Waiting

zach030 added the protocol integration label Oct 13, 2026

feat(lunarbase): enhance quote limit calculation and add tests for mi…

c8273e9

…nimum trade limits

ThomasAqu1nas requested a deployment to integration-tests October 13, 2026 11:21 — with GitHub Actions Waiting

zach030 reviewed Oct 13, 2026

View reviewed changes

Navigation Menu

Repository navigation

feat(lunarbase): support v0.4.1 pools and BSC indexing - #1445

feat(lunarbase): support v0.4.1 pools and BSC indexing#1445
ThomasAqu1nas wants to merge 4 commits into
propeller-heads:mainfrom
Lunarbase-Lab:v0.4.1

ThomasAqu1nas commented Sep 15, 2026

Uh oh!

socket-security Bot commented Sep 15, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

zach030 Sep 23, 2026

Uh oh!

zach030 Oct 6, 2026

Uh oh!

Uh oh!

github-actions Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

ThomasAqu1nas commented Oct 5, 2026

Uh oh!

zach030 Oct 6, 2026 •

edited

Loading

Uh oh!

zach030 Oct 13, 2026

Uh oh!

zach030 Oct 13, 2026

Uh oh!

This branch is waiting to be deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Direction	Anchor	Input reserve	`quote_exact_in(3)` output	`get_limits`
X → Y	`3 * Q96 / 4`	0	1	`(0, 0)`
X → Y	`3 * Q96 / 4`	1	1	`(0, 0)`
Y → X	`floor(4 * Q96 / 3)`	0	1	`(0, 0)`
Y → X	`floor(4 * Q96 / 3)`	1	1	`(0, 0)`

		# This historical range predates router whitelisting. Execution comparison
		# requires a whitelisted router; recapture a later range for on-chain parity.

Direction / input	On-chain quote	Current simulator
X → Y / 0.001 ETH	2,390,779	2,511,757
Y → X / 1 USDC	393,595,800,163,007	397,890,571,043,066

Conversation

ThomasAqu1nas commented Sep 15, 2026

Uh oh!

socket-security Bot commented Sep 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

zach030 Sep 23, 2026

Choose a reason for hiding this comment

Uh oh!

zach030 Oct 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ThomasAqu1nas commented Oct 5, 2026

Uh oh!

zach030 Oct 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zach030 Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

zach030 Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

This branch is waiting to be deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

socket-security Bot commented Sep 15, 2026 •

edited

Loading

github-actions Bot commented Oct 5, 2026 •

edited

Loading

zach030 Oct 6, 2026 •

edited

Loading