feat: escrow account sweeping by keruch · Pull Request #220 · dymensionxyz/hyperlane-monorepo

keruch · 2025-08-08T21:08:14Z

#214

keruch · 2025-08-11T16:53:07Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

there was a bug when messages was overwriting itself

keruch · 2025-08-11T16:53:47Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

I removed must_match and used escrow_public + address_prefix so I can easily use validate_pskts in the demo

keruch · 2025-08-11T16:54:18Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

Important part!

keruch · 2025-08-12T12:35:57Z

dymension/libs/kaspa/demo/multisig/src/main.rs

I added validation to the demo (without any hub call though)

keruch · 2025-08-12T12:39:37Z

dymension/libs/kaspa/lib/relayer/src/withdraw/messages.rs

I created this function to build WithdrawFXG without any hub calls

keruch · 2025-08-12T18:21:04Z

dymension/libs/kaspa/lib/relayer/src/withdraw/hub_to_kaspa.rs

+    .map_err(|e| eyre::eyre!("Estimate TX mass: {e}"))?;
+
+    // Apply TX mass multiplier and feerate
+    let tx_fee = (tx_mass as f64 * TX_MASS_MULTIPLIER * feerate).round() as u64;


I added feerate, it closes #112

keruch · 2025-08-12T18:23:56Z

dymension/libs/kaspa/lib/relayer/src/withdraw/hub_to_kaspa.rs

+    Ok((anchor_input, escrow_inputs))
+}
+
+fn estimate_mass(


I updated estimate_fee (now it is estimate_mass) to use a different MassCalculator which kaspa is using in the tx generator. I found it more accurate and I stopped getting insufficient fee errors.

keruch · 2025-08-12T18:28:03Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

+    let tx_id = pskt.calculate_id();
+
+    // If there are no messages and payload is empty, then the PSKT
+    // is a sweeping tx which does not spend the anchor
+    let tx_type = if expected_messages.is_empty() && pskt.global.payload.is_none() {
+        info!("PSKT is a sweeping tx: {tx_id}");
+        TxType::Sweeping
+    } else {
+        info!("PSKT is a withdrawal tx: {tx_id}");
+        TxType::Withdrawal
+    };
+
+    let ix = validate_pskt_application_semantics(
+        &pskt,
+        must_spend,
+        tx_type,
+        expected_messages,
+        escrow_public,
+        address_prefix,
+    )?;
+
+    match tx_type {
+        // In case of the sweeping tx, try to spend the anchor in the next PSKT
+        TxType::Sweeping => Ok(must_spend),
+        TxType::Withdrawal => Ok(TransactionOutpoint::new(pskt.calculate_id(), ix)),
+    }


important to check!!

keruch · 2025-08-12T18:30:14Z

dymension/tests/kaspa_hub_test_kas/agent-config.json

This config worked for me, maybe worth copying from https://github.com/dymensionxyz/hyperlane-deployments/blob/main/playground/guide_kaspa_kas/roles/relayer/agent-config.json

danwt · 2025-08-13T08:36:38Z

dymension/libs/kaspa/demo/multisig/src/main.rs

-    let args = Args::parse();
+    let mut args = Args::parse();
+
+    args.wallet_secret = Some("123456qwe".to_string());


security is my second name

danwt

Great work! Just a few clarifying questions

danwt · 2025-08-13T08:45:12Z

dymension/libs/kaspa/lib/core/src/balance.rs

+    let utxos = rpc
+        .get_utxos_by_addresses(vec![addr.clone()])
        .await
-        .map_err(|e| Error::Custom(format!("Getting balance for address: {e}")))?;
+        .map_err(|e| Error::Custom(format!("Getting UTXOs for address: {e}")))?;
+
+    let num = utxos.len();
+    let balance: u64 = utxos.into_iter().map(|u| u.utxo_entry.amount).sum();
+
+    info!("{} has {} UTXOs and {} balance", source, num, balance);

-    info!("{} balance: {}", source, balance);
    Ok(balance)


@keruch what was wrong with the old way?

nothing, I just wanted to know the number of UTXOs

danwt · 2025-08-13T09:05:04Z

dymension/libs/kaspa/lib/hardcode/src/tx.rs

+// Add this small priproty fee to every sweeping TX to ensure the sufficient fee even if there
+// is one validator. 3_000 is a magic number.
+// TODO: make it configurable?
+pub const RELAYER_SWEEPING_PRIORITY_FEE: u64 = 3_000;


is this also going to work on mainnet?

not sure, that's why I think it's worth making it configurable. 3000 is just what I came up with while testing.

danwt · 2025-08-13T09:10:35Z

rust/main/hyperlane-base/src/kas_hack/logic_loop.rs

+        let tx_hash = h512_to_cosmos_hash(outcome.transaction_id).encode_hex_upper::<String>();
+
+        if !outcome.executed {
+            return Err(eyre::eyre!(
+                "Indicate progress failed, TX was not executed on-chain, tx hash: {tx_hash}"
+            ));
+        }
+
        info!(
-            "Dymension, indicated progress on hub: {:?}, outcome: {:?}",
-            fxg.progress_indication, outcome
+            "Dymension, indicated progress on hub: {:?}, outcome: {:?}, tx hash: {:?}",
+            fxg.progress_indication, outcome, tx_hash,
        );


danwt · 2025-08-13T09:34:28Z

dymension/libs/kaspa/lib/relayer/src/withdraw/messages.rs

+    let messages = {
+        let sweep_count = sweeping_bundle.as_ref().map_or(0, |b| b.0.len());
+        let mut messages = Vec::with_capacity(sweep_count + valid_msgs.len());
+        messages.extend(vec![Vec::new(); sweep_count]);


what is the Vec::new() here for?

vec![V; N] (with semicolon) is a macro to create a vector of V repeated N times. so with vec![Vec::new(); sweep_count] I create a vector of sweep_count empty vectors, ie [[], [], [], ...].

danwt · 2025-08-13T10:01:14Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

+        _ => {}
    }

    let payload_expect = MessageIDs::from(expected_messages).to_bytes();


the payload check still works properly? i.e. sweep is checked to be empty?

yes! it's here

hyperlane-monorepo/dymension/libs/kaspa/lib/validator/src/withdraw.rs

Lines 274 to 280 in 78be165

let tx_type = if expected_messages.is_empty() && pskt.global.payload.is_none() {

info!("PSKT is a sweeping tx: {tx_id}");

TxType::Sweeping

} else {

info!("PSKT is a withdrawal tx: {tx_id}");

TxType::Withdrawal

};

danwt · 2025-08-13T10:06:20Z

dymension/libs/kaspa/lib/validator/src/withdraw.rs

+
+    // If there are no messages and payload is empty, then the PSKT
+    // is a sweeping tx which does not spend the anchor
+    let tx_type = if expected_messages.is_empty() && pskt.global.payload.is_none() {


is this clause strictly needed?

&& pskt.global.payload.is_none()
it's confusing me

i think the code will still be correct without this?

I wanted to distinguish between sweeping and withdrawal TXs in the validation. In case of sweeping, we don't want the anchor to be an input. In case of withdrawal, we do.

hyperlane-monorepo/dymension/libs/kaspa/lib/validator/src/withdraw.rs

Lines 329 to 340 in 78be165

// Check that the PSKT is a sweeping tx or a withdrawal tx.

// If it is a sweeping tx, then the anchor must not be spent.

// If it is a withdrawal tx, then the anchor must be spent.

match (tx_type, anchor_found) {

(TxType::Sweeping, true) => {

return Err(ValidationError::AnchorSpent { o: current_anchor });

}

(TxType::Withdrawal, false) => {

return Err(ValidationError::AnchorNotFound { o: current_anchor });

}

_ => {}

}

Right, but I mean, you can check expected messages empty, there is no need to check the payload too in this if statement?
It makes it harder to understand IMO because then there is the payload correctness check in the next function clal

the payload correctness check in the next function

now I got it! fixed

- Changed PopulatedInput to include optional redeem_script as third element - Keep signature_script empty for unsigned transactions (as it should be) - Store redeem_script separately and only add to PSKT when needed - Update all usages to handle the new tuple structure This fixes the semantic confusion where redeem_script was incorrectly stored in the signature_script field of TransactionInput.

dymension/libs/kaspa/lib/validator/src/withdraw.rs

keruch self-assigned this Aug 8, 2025

keruch commented Aug 11, 2025

View reviewed changes

dymension/libs/kaspa/lib/validator/src/withdraw.rs Outdated

Comment on lines 333 to 339

Copy link
Copy Markdown

Author

keruch Aug 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Important part!

keruch commented Aug 12, 2025

View reviewed changes

keruch force-pushed the kirill/sweeping branch from a975f39 to b7e36bc Compare August 12, 2025 18:18

keruch commented Aug 12, 2025

View reviewed changes

keruch marked this pull request as ready for review August 12, 2025 18:32

danwt reviewed Aug 13, 2025

View reviewed changes

danwt previously approved these changes Aug 13, 2025

View reviewed changes

keruch dismissed danwt’s stale review via 37ac1e7 August 13, 2025 10:37

danwt self-requested a review August 13, 2025 11:19

keruch and others added 14 commits August 13, 2025 13:05

feat: escrow account sweeping

cc9f9a8

feat: escrow account sweeping

13b3b06

feat: escrow account sweeping 1

91868e4

cargo fix

6eca00f

cargo fix

96422ef

sweeping threshold

d5b8629

cargo fix

87cb740

minors

fd59916

fmt

9331cdf

update claude.md

fff2c87

fix demo wallet password

01c8405

progress checkpoint: code might be broken

4858cbb

claude

6792160

typo

d30796f

danwt and others added 6 commits August 13, 2025 13:05

comment

346d39e

clarification

eeb1438

typos

4d616d0

fmt

274fab8

threads

ed3d3d6

danwt force-pushed the kirill/sweeping branch from 1c2db1a to 27ffc02 Compare August 13, 2025 12:17

srene reviewed Aug 14, 2025

View reviewed changes

dymension/libs/kaspa/lib/validator/src/withdraw.rs Show resolved Hide resolved

srene approved these changes Aug 14, 2025

View reviewed changes

danwt approved these changes Aug 14, 2025

View reviewed changes

danwt merged commit 5af7f5a into main-dym Aug 14, 2025
9 of 62 checks passed

keruch mentioned this pull request Aug 14, 2025

TX mass constraint optimization #214

Closed

	let tx_type = if expected_messages.is_empty() && pskt.global.payload.is_none() {
	info!("PSKT is a sweeping tx: {tx_id}");
	TxType::Sweeping
	} else {
	info!("PSKT is a withdrawal tx: {tx_id}");
	TxType::Withdrawal
	};

	// Check that the PSKT is a sweeping tx or a withdrawal tx.
	// If it is a sweeping tx, then the anchor must not be spent.
	// If it is a withdrawal tx, then the anchor must be spent.
	match (tx_type, anchor_found) {
	(TxType::Sweeping, true) => {
	return Err(ValidationError::AnchorSpent { o: current_anchor });
	}
	(TxType::Withdrawal, false) => {
	return Err(ValidationError::AnchorNotFound { o: current_anchor });
	}
	_ => {}
	}

Conversation

keruch commented Aug 8, 2025 • edited by danwt Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

danwt left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

keruch commented Aug 8, 2025 •

edited by danwt

Loading