feat(rpc): greatly enhance `simulate` RPC by mvadari · Pull Request #5637 · XRPLF/rippled

mvadari · 2025-07-31T13:50:08Z

High Level Overview of Change

Currently, the simulate RPC only supports simulating transactions in the current ledger. This PR adds support to also simulate transactions in past ledgers, as long as the rippled node has that data.

This will make it much easier to debug past transactions, without the difficulty/complexity of needing to perform a full ledger replay (although it doesn't quite have the same flexibility). It'll provide about 90-95% of the functionality of ledger replay (if not 100%) for like 1% of the effort.

Currently, you need to follow a long set of instructions with complex scripts to run ledger replay. With this feature, the process for "replaying" a transaction would be as follows:

Set a breakpoint in gdb/lldb/etc in whatever transactor you're investigating
Fetch the n-1 and n ledger that the transaction you're investigating is in (steps 1 and 2 are somewhat interchangeable)
Use the simulate RPC to simulate the past transaction in the past ledger as described below:

{
  "command": "simulate",
  "tx_hash": <insert tx hash here>,
  "ledger_index": <n>
}

If you need to include multiple transactions (i.e. there were other relevant transactions in the same ledger):

{
  "command": "simulate",
  "transactions": [
    {
      "tx_hash": <insert tx hash here>
    }
    ...
  ],
  "ledger_index": <n>
}

Depending on performance testing results, we may want this specific feature to be admin-only.

New features:

Can simulate transactions on past ledgers
Can fetch transactions from past ledgers (only works if you're also executing on past ledgers)
Batch processing
Process many transactions in the same request
TODO: toggle amendments

Context of Change

Closes #5540 (probably need to still add amendment support for that)

Type of Change

New feature (non-breaking change which adds functionality)

API Impact

Public API: New feature (new methods and/or new fields)
Public API: Breaking change (in general, breaking changes should only impact the next api_version)
libxrpl change (any change that may affect libxrpl or dependents of libxrpl)
Peer protocol change (must be backward compatible or bump the peer protocol version)

Before / After

Mainly changes to the simulate RPC. Some error codes change on other RPCs, but they're pretty minor.

Test Plan

Added some basic tests

Future Tasks

See TODOs above
More tests
Better autofilling support for Batch/etc

codecov · 2025-07-31T14:22:36Z

Codecov Report

❌ Patch coverage is 97.53695% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 79.8%. Comparing base (afc660a) to head (2836a7d).

Files with missing lines	Patch %	Lines
src/xrpld/rpc/handlers/Simulate.cpp	97.8%	4 Missing ⚠️
src/libxrpl/tx/apply.cpp	94.1%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           develop   #5637    +/-   ##
========================================
  Coverage     79.8%   79.8%            
========================================
  Files          858     858            
  Lines        67757   67888   +131     
  Branches      7557    7560     +3     
========================================
+ Hits         54064   54190   +126     
- Misses       13693   13698     +5

Files with missing lines	Coverage Δ
src/libxrpl/ledger/ApplyStateTable.cpp	`92.7% <100.0%> (-<0.1%)`	⬇️
src/xrpld/rpc/handlers/LedgerData.cpp	`56.2% <100.0%> (ø)`
src/libxrpl/tx/apply.cpp	`93.5% <94.1%> (-0.6%)`	⬇️
src/xrpld/rpc/handlers/Simulate.cpp	`98.5% <97.8%> (-1.5%)`	⬇️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…t break anything)

Bronek · 2025-08-08T12:59:07Z

Set a breakpoint in gdb/lldb in whatever transactor you're investigating

Reminder that Transactor::trapTransaction was specifically created to provide a convenient location for such breakpoint, in case if you do not use simulate but replay the whole ledger instead (which is obviously more trouble that the alternative process proposed here)

rippled/src/xrpld/app/tx/detail/Transactor.cpp

Line 1054 in 39b5031

Transactor::trapTransaction(uint256 txHash) const

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 5 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-13T22:16:34Z

src/xrpld/rpc/handlers/Simulate.cpp

+         * cannot use `simulate` to bypass signature checks and submit
+         * transactions/modify the current ledger directly.
+         ***************************************/
+        auto const result = apply(context.app, perTxView, txn, tapDRY_RUN, context.j);


Simulation now uses the low-level apply(...) call unconditionally. For the current/open ledger this bypasses TxQ::apply behavior (fee escalation, queuing decisions, and related local checks), so simulate results may diverge from what would happen when actually submitting a transaction under load. If the intent is to match submit semantics on the current ledger, consider using context.app.getTxQ().apply(..., tapDRY_RUN, ...) when isCurrentLedger is true, and only use apply(...) for historical-ledger simulation.

Suggested change

auto const result = apply(context.app, perTxView, txn, tapDRY_RUN, context.j);

ApplyResult const result =

isCurrentLedger

? context.app.getTxQ().apply(

context.app, perTxView, txn, tapDRY_RUN, context.j)

: apply(context.app, perTxView, txn, tapDRY_RUN, context.j);

Copilot · 2026-02-13T22:16:34Z

src/xrpld/rpc/handlers/Simulate.cpp

    if (tx_json.isMember(jss::error))
-        return tx_json;
+        return Unexpected(tx_json);



For multi-transaction requests, autofill (especially Sequence) is performed in processTransaction using lpLedger before any earlier simulated transactions are applied. This can cause later transactions in the same request to be autofilled with the same Sequence and then fail (the test currently works around this by manually setting the second Sequence). Consider either (1) moving autofill into the simulation loop so it can consult the evolving view, or (2) requiring clients to provide explicit Sequence values when transactions contains more than one entry.

Suggested change

// If this is a multi-transaction simulate request, require that each

// transaction explicitly specify Sequence or TicketSequence so that

// we do not autofill potentially conflicting values from the same

// base ledger view for multiple transactions.

bool const isMultiTxRequest =

context.params.isMember(jss::transactions) &&

context.params[jss::transactions].isArray() &&

context.params[jss::transactions].size() > 1;

if (isMultiTxRequest)

{

bool const hasTicketSeq = tx_json.isMember(sfTicketSequence.jsonName);

bool const hasSeq = tx_json.isMember(sfSequence.jsonName);

if (!hasTicketSeq && !hasSeq)

{

Json::Value jvResult = Json::objectValue;

jvResult[jss::error] = rpcINVALID_PARAMS;

jvResult[jss::error_message] =

"When submitting multiple transactions in a single simulate "

"request, each transaction must specify either Sequence or "

"TicketSequence.";

return Unexpected(jvResult);

}

}

Copilot · 2026-02-13T22:16:34Z

src/xrpld/rpc/handlers/Simulate.cpp

 {
-    context.loadType = Resource::feeMediumBurdenRPC;
+    Json::Value jvTransactions = Json::arrayValue;
+    std::vector<ApplyResult> results;


simulateTxn declares std::vector<ApplyResult> results; but never uses it. Removing the unused variable would reduce noise and avoid confusion about whether results are meant to be collected.

Suggested change

std::vector<ApplyResult> results;

Copilot · 2026-02-13T22:16:35Z

src/xrpld/rpc/handlers/Simulate.cpp

+#include <xrpld/app/ledger/TransactionMaster.h>
+#include <xrpld/app/misc/HashRouter.h>
 #include <xrpld/app/misc/Transaction.h>
-#include <xrpld/app/misc/TxQ.h>
 #include <xrpld/app/tx/apply.h>


The added HashRouter includes (<xrpld/app/misc/HashRouter.h> / <xrpl/core/HashRouter.h>) don't appear to be used anywhere in this file. If they aren't required for compilation, consider removing them to keep dependencies minimal.

Copilot · 2026-02-13T22:16:35Z

src/xrpld/rpc/handlers/Simulate.cpp

    else
    {
-        return RPC::make_param_error("Neither `tx_blob` nor `tx_json` included.");
+        auto const result = getTxJsonFromHistory(context, isCurrentLedger);
+        if (result.isMember(jss::error))
+        {
+            return result;
+        }
+        else
+        {
+            tx_json = result;
+        }
    }


getTxJsonFromParams falls back to getTxJsonFromHistory(context, ...), but getTxJsonFromHistory reads context.params instead of the per-transaction txInput. In a batch request (transactions array), items that specify tx_hash or ctid will therefore not be read correctly (and may end up fetching using a default/zero hash). Consider passing the per-item JSON into getTxJsonFromHistory (or parsing tx_hash/ctid directly from txInput) so batch history lookups work as intended.

…-past

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 10 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-24T22:28:12Z

src/xrpld/rpc/handlers/Simulate.cpp

+                    STTx const txn(
+                        static_cast<TxType>(rawTxn.getFieldU16(sfTransactionType)),
+                        [&rawTxn](STObject& obj) { obj = STObject(rawTxn); });
+                    jvTransactions.append(
+                        processResult(innerResult, txn, isBinaryOutput, wholeBatchView));


Variable shadowing: the inner txn variable declared here shadows the outer txn variable from line 402. This can lead to confusion and potential bugs. Consider renaming the inner variable to something like innerTxn or batchInnerTxn to make the distinction clear.

Suggested change

STTx const txn(

static_cast<TxType>(rawTxn.getFieldU16(sfTransactionType)),

[&rawTxn](STObject& obj) { obj = STObject(rawTxn); });

jvTransactions.append(

processResult(innerResult, txn, isBinaryOutput, wholeBatchView));

STTx const innerTxn(

static_cast<TxType>(rawTxn.getFieldU16(sfTransactionType)),

[&rawTxn](STObject& obj) { obj = STObject(rawTxn); });

jvTransactions.append(

processResult(innerResult, innerTxn, isBinaryOutput, wholeBatchView));

Copilot · 2026-02-24T22:28:12Z

src/xrpld/rpc/handlers/Simulate.cpp

+                context.registry.config(),
+                context.registry.getFeeTrack(),
+                context.registry.getTxQ(),
+                context.registry,


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. All occurrences of context.registry in this function (lines 141-144) should be changed to context.app.

Copilot · 2026-02-24T22:28:13Z

src/xrpld/rpc/handlers/Simulate.cpp

    if (!tx_json.isMember(jss::NetworkID))
    {
-        auto const networkId = context.app.getNetworkIDService().getNetworkID();
+        auto const networkId = context.registry.getNetworkIDService().getNetworkID();


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. This should be context.app.getNetworkIDService() instead of context.registry.getNetworkIDService().

Suggested change

auto const networkId = context.registry.getNetworkIDService().getNetworkID();

auto const networkId = context.app.getNetworkIDService().getNetworkID();

Copilot · 2026-02-24T22:28:13Z

src/xrpld/rpc/handlers/Simulate.cpp

+        if (auto const optHash = context.registry.getLedgerMaster().txnIdFromIndex(ledgerSq, txId);
+            optHash)
+        {
+            hash = *optHash;
+        }
+        else
+        {
+            return RPC::make_error(rpcTXN_NOT_FOUND);
+        }
+    }
+    using TxPair = std::pair<std::shared_ptr<Transaction>, std::shared_ptr<TxMeta>>;
+    auto ec{rpcSUCCESS};
+    std::variant<TxPair, TxSearched> v = context.registry.getMasterTransaction().fetch(hash, ec);


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. All occurrences of context.registry in this function (lines 217, 229) should be changed to context.app.

Copilot · 2026-02-24T22:28:13Z

src/xrpld/rpc/handlers/Simulate.cpp

+        auto const result = apply(context.registry, perTxView, txn, tapDRY_RUN, context.j);
+        if (isTesSuccess(result.ter) || isTecClaim(result.ter))
+            perTxView.apply(view);
+        jvTransactions.append(processResult(result, txn, isBinaryOutput, view));
+
+        if (isTesSuccess(result.ter) && txn.getTxnType() == ttBATCH)
+        {
+            OpenView wholeBatchView(batch_view, view);

-    // check validity of `binary` param
-    if (context.params.isMember(jss::binary) && !context.params[jss::binary].isBool())
+            if (auto const batchResults = applyBatchTransactions(
+                    context.registry, wholeBatchView, txn, tapDRY_RUN, context.j);


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. All occurrences of context.registry in this function (lines 411, 421) should be changed to context.app.

Copilot · 2026-02-24T22:28:13Z

include/xrpl/tx/apply.h

+std::optional<std::vector<ApplyResult>>
+applyBatchTransactions(
+    Application& app,
+    OpenView& batchView,
+    STTx const& batchTxn,
+    ApplyFlags flags,
+    beast::Journal j);


The function signature in the header declares Application& app as the first parameter, but the implementation in src/libxrpl/tx/apply.cpp (line 140) uses ServiceRegistry& registry. These must match. Since Application inherits from ServiceRegistry and the implementation uses ServiceRegistry, the header should be updated to use ServiceRegistry& registry for consistency with the implementation.

Copilot · 2026-02-24T22:28:14Z

src/xrpld/rpc/handlers/Simulate.cpp

+                    context.registry, wholeBatchView, txn, tapDRY_RUN, context.j);
+                batchResults)
+            {
+                for (int i = 0; i < batchResults->size(); ++i)


Using int for the loop counter when iterating over batchResults->size() could cause signed/unsigned comparison warnings or issues. Use std::size_t or an unsigned type instead, consistent with the return type of size().

Suggested change

for (int i = 0; i < batchResults->size(); ++i)

for (std::size_t i = 0; i < batchResults->size(); ++i)

Copilot · 2026-02-24T22:28:14Z

src/xrpld/rpc/handlers/Simulate.cpp

+    if (!sle)
    {
-        JLOG(context.app.journal("Simulate").debug())
+        JLOG(context.registry.journal("Simulate").debug())


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. This should be context.app.journal("Simulate").debug() instead of context.registry.journal("Simulate").debug().

Copilot · 2026-02-24T22:28:14Z

src/xrpld/rpc/handlers/Simulate.cpp

+        return sle->getFieldU32(sfSequence);

-    return hasTicketSeq ? 0 : context.app.getTxQ().nextQueuableSeq(sle).value();
+    return context.registry.getTxQ().nextQueuableSeq(sle).value();


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. All occurrences of context.registry in this function should be changed to context.app.

Copilot · 2026-02-24T22:28:14Z

src/xrpld/rpc/handlers/Simulate.cpp


-    if (stTx->getTxnType() == ttBATCH)
+    std::string reason;
+    return std::make_shared<Transaction>(stTx, reason, context.registry);


The RPC::JsonContext struct does not have a registry member. Based on the Context.h file, the member is named app. This should be context.app instead of context.registry.

Suggested change

return std::make_shared<Transaction>(stTx, reason, context.registry);

return std::make_shared<Transaction>(stTx, reason, context.app);

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-26T17:20:31Z

include/xrpl/tx/apply.h


+std::optional<std::vector<ApplyResult>>
+applyBatchTransactions(
+    Registry& registry,


The parameter type is declared as Registry& but should be ServiceRegistry& to match the implementation in apply.cpp. The type Registry is not defined in the codebase and this will cause a compilation error.

Suggested change

Registry& registry,

ServiceRegistry& registry,

Copilot · 2026-02-26T17:20:32Z

src/xrpld/rpc/handlers/Simulate.cpp

+    if (!isCurrentLedger)
+        return sle->getFieldU32(sfSequence);

-    return hasTicketSeq ? 0 : context.app.getTxQ().nextQueuableSeq(sle).value();
+    return context.registry.getTxQ().nextQueuableSeq(sle).value();


When simulating multiple transactions in the same request, the sequence autofilling doesn't automatically increment for subsequent transactions. Users must manually specify the sequence for all transactions after the first one. This should be documented or fixed to automatically increment sequences for chained transactions within the same simulation batch.

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 12 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-26T20:07:18Z

src/libxrpl/tx/apply.cpp


-        if (result.applied)
-            ++applied;
+        if (result.applied || flags & tapDRY_RUN)


For better readability, consider adding explicit parentheses around the bitwise AND operation: (result.applied || (flags & tapDRY_RUN)). While the operator precedence is correct, the explicit parentheses make the intent clearer and prevent potential confusion.

Suggested change

if (result.applied || flags & tapDRY_RUN)

if (result.applied || (flags & tapDRY_RUN))

Copilot · 2026-02-26T20:07:18Z

src/xrpld/rpc/handlers/Simulate.cpp

    if (!tx_json.isMember(jss::NetworkID))
    {
-        auto const networkId = context.app.getNetworkIDService().getNetworkID();
+        auto const networkId = context.registry.getNetworkIDService().getNetworkID();


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:18Z

src/xrpld/rpc/handlers/Simulate.cpp

+         * cannot use `simulate` to bypass signature checks and submit
+         * transactions/modify the current ledger directly.
+         ***************************************/
+        auto const result = apply(context.registry, perTxView, txn, tapDRY_RUN, context.j);


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:18Z

src/xrpld/rpc/handlers/Simulate.cpp

-    // check validity of `binary` param
-    if (context.params.isMember(jss::binary) && !context.params[jss::binary].isBool())
+            if (auto const batchResults = applyBatchTransactions(
+                    context.registry, wholeBatchView, txn, tapDRY_RUN, context.j);


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:19Z

src/libxrpl/ledger/ApplyStateTable.cpp

+    to.rawTxInsert(tx.getTransactionID(), sTx, sMeta);
+    apply(to);


Removing the isDryRun check means transactions are now inserted into the view even during dry runs. While this should be safe because simulate uses temporary OpenView copies, this changes the behavior where previously dry runs would not insert transactions into views at all. Consider verifying this behavior change is intentional and document it if necessary, as it may have implications for how views track transaction insertions during simulation.

Suggested change

to.rawTxInsert(tx.getTransactionID(), sTx, sMeta);

apply(to);

if (!isDryRun)

{

to.rawTxInsert(tx.getTransactionID(), sTx, sMeta);

apply(to);

}

Copilot · 2026-02-26T20:07:19Z

src/xrpld/rpc/handlers/Simulate.cpp

+                context.registry.config(),
+                context.registry.getFeeTrack(),
+                context.registry.getTxQ(),
+                context.registry,


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:20Z

src/xrpld/rpc/handlers/Simulate.cpp

+            return RPC::invalid_field_error(jss::ctid);
+        }
+        auto const [ledgerSq, txId, _] = *decodedCTID;
+        if (auto const optHash = context.registry.getLedgerMaster().txnIdFromIndex(ledgerSq, txId);


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:20Z

src/xrpld/rpc/handlers/Simulate.cpp

+    }
+    using TxPair = std::pair<std::shared_ptr<Transaction>, std::shared_ptr<TxMeta>>;
+    auto ec{rpcSUCCESS};
+    std::variant<TxPair, TxSearched> v = context.registry.getMasterTransaction().fetch(hash, ec);


The code uses context.registry but should use context.app. The RPC::JsonContext struct only has a member named app, not registry.

Copilot · 2026-02-26T20:07:20Z

src/xrpld/rpc/handlers/Simulate.cpp

+    if (!sle)
    {
-        JLOG(context.app.journal("Simulate").debug())
+        JLOG(context.registry.journal("Simulate").debug())


The code uses context.registry but the RPC::JsonContext struct only has a member named app, not registry. This will cause a compilation error. All occurrences of context.registry throughout this file should be changed to context.app. The Application class inherits from ServiceRegistry, so context.app can be used wherever a ServiceRegistry reference is needed.

Copilot · 2026-02-26T20:07:20Z

src/xrpld/rpc/handlers/Simulate.cpp

+    context.loadType = Resource::feeMediumBurdenRPC;
+


The resource load type is set to feeMediumBurdenRPC regardless of the number of transactions being simulated. Simulating 1000 transactions (the maximum allowed) will consume significantly more resources than simulating a single transaction, but both are charged the same. Consider adjusting the load type based on the number of transactions, or using a higher burden level like feeHighBurdenRPC for the simulate RPC to account for its potentially expensive operations.

Suggested change

context.loadType = Resource::feeMediumBurdenRPC;

// Set load type based on how many transactions are being simulated.

std::size_t numSimulatedTxs = 0;

if (context.params.isMember(jss::transactions) &&

context.params[jss::transactions].isArray())

{

numSimulatedTxs = context.params[jss::transactions].size();

}

else

{

// At most one of tx_json, tx_blob, tx_hash, or ctid is expected.

// If none are present, numSimulatedTxs remains 0 and the validation

// below will return an error.

numSimulatedTxs = 1;

}

context.loadType = numSimulatedTxs > 1

? Resource::feeHighBurdenRPC

: Resource::feeMediumBurdenRPC;

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 5 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-27T18:55:02Z

src/xrpld/rpc/handlers/Simulate.cpp

+        if (isCurrentLedger)
+        {
+            return RPC::make_param_error(
+                "Cannot use `tx_hash` without `ledger_index` or "
+                "`ledger_hash`.");
+        }


The error message returned when isCurrentLedger is true says "Cannot use tx_hash without ledger_index or ledger_hash." If the caller explicitly provides ledger_index: "current", this message is misleading (they did provide ledger_index). Consider rewording to indicate that tx_hash/ctid are only supported when simulating against a non-current (closed/validated/historical) ledger.

Copilot · 2026-02-27T18:55:03Z

src/xrpld/rpc/handlers/Simulate.cpp

        JLOG(context.app.journal("Simulate").debug())
            << "Failed to find source account "
            << "in current ledger: " << toBase58(*srcAddressID);

        return Unexpected(rpcError(rpcSRC_ACT_NOT_FOUND));


This debug log message hard-codes "in current ledger", but getAutofillSequence can now be called with a past ledger view (lpLedger). Consider adjusting the text (or including the ledger seq) so logs remain accurate when simulating historical ledgers.

Copilot · 2026-02-27T18:55:03Z

src/xrpld/rpc/handlers/Simulate.cpp

@@ -178,17 +281,25 @@ getTxJsonFromParams(Json::Value const& params)
            return RPC::invalid_field_error(jss::tx_blob);
        }
    }
-    else if (params.isMember(jss::tx_json))
+    else if (txInput.isMember(jss::tx_json))
    {
-        tx_json = params[jss::tx_json];
+        tx_json = txInput[jss::tx_json];
        if (!tx_json.isObject())
        {
            return RPC::object_field_error(jss::tx_json);
        }
    }
    else
    {
-        return RPC::make_param_error("Neither `tx_blob` nor `tx_json` included.");
+        auto const result = getTxJsonFromHistory(context, isCurrentLedger);
+        if (result.isMember(jss::error))
+        {
+            return result;
+        }
+        else
+        {
+            tx_json = result;
+        }
    }


getTxJsonFromParams only checks tx_blob / tx_json. For per-item objects in the transactions array that specify tx_hash or ctid, it falls into the history path, but getTxJsonFromHistory reads from context.params (top-level) instead of the per-item object. As a result, batch items cannot actually be sourced by tx_hash/ctid despite being accepted by validation. Consider passing the per-transaction object into getTxJsonFromHistory (or parsing tx_hash/ctid directly in getTxJsonFromParams) so batch items work as documented.

Copilot · 2026-02-27T18:55:03Z

src/xrpld/rpc/handlers/Simulate.cpp

 {
-    context.loadType = Resource::feeMediumBurdenRPC;
+    Json::Value jvTransactions = Json::arrayValue;
+    std::vector<ApplyResult> results;


simulateTxn declares std::vector<ApplyResult> results; but never uses it. With -Wextra (and often -Werror) this can fail the build due to an unused-variable warning. Remove it or use it.

Suggested change

std::vector<ApplyResult> results;

Copilot · 2026-02-27T18:55:03Z

src/xrpld/rpc/handlers/Simulate.cpp

+         * cannot use `simulate` to bypass signature checks and submit
+         * transactions/modify the current ledger directly.
+         ***************************************/
+        auto const result = apply(context.app, perTxView, txn, tapDRY_RUN, context.j);


For simulations against the current open ledger, this switched from TxQ::apply to xrpl::apply(...). That bypasses TxQ logic (fee escalation / queuing constraints) and can change engine_result compared to what would happen on actual submission. Consider using context.app.getTxQ().apply(...) when isCurrentLedger is true, and only falling back to xrpl::apply(...) for non-current ledgers.

Suggested change

auto const result = apply(context.app, perTxView, txn, tapDRY_RUN, context.j);

ApplyResult result;

if (isCurrentLedger)

result = context.app.getTxQ().apply(perTxView, txn, tapDRY_RUN, context.j);

else

result = apply(context.app, perTxView, txn, tapDRY_RUN, context.j);

Copilot

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 6 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-02T22:13:40Z

src/xrpld/rpc/handlers/Simulate.cpp

+getTxJsonFromHistory(RPC::JsonContext& context, bool const isCurrentLedger)
 {
-    Json::Value tx_json;
+    auto const params = context.params;
+    uint256 hash;
+    if (params.isMember(jss::tx_hash))
+    {
+        auto const tx_hash = params[jss::tx_hash];
+        if (!tx_hash.isString())
+        {
+            return RPC::invalid_field_error(jss::tx_hash);
+        }
+        if (isCurrentLedger)
+        {
+            return RPC::make_param_error(
+                "Cannot use `tx_hash` without `ledger_index` or "
+                "`ledger_hash`.");
+        }
+        if (!hash.parseHex(context.params[jss::tx_hash].asString()))
+            return RPC::invalid_field_error(jss::tx_hash);
+    }
+    else if (params.isMember(jss::ctid))
+    {
+        auto const ctid = params[jss::ctid];
+        if (!ctid.isString())
+        {
+            return RPC::invalid_field_error(jss::ctid);
+        }
+        if (isCurrentLedger)
+        {
+            return RPC::make_param_error(
+                "Cannot use `ctid` without `ledger_index` or `ledger_hash`.");
+        }
+        auto decodedCTID = RPC::decodeCTID(context.params[jss::ctid].asString());
+        if (!decodedCTID)
+        {
+            return RPC::invalid_field_error(jss::ctid);
+        }
+        auto const [ledgerSq, txId, _] = *decodedCTID;
+        if (auto const optHash = context.app.getLedgerMaster().txnIdFromIndex(ledgerSq, txId);
+            optHash)
+        {
+            hash = *optHash;
+        }
+        else
+        {
+            return RPC::make_error(rpcTXN_NOT_FOUND);
+        }
+    }


getTxJsonFromParams(context, txInput, ...) falls back to getTxJsonFromHistory(context, ...), but getTxJsonFromHistory only reads context.params (top-level request), not the per-item txInput. As a result, batch items that specify tx_hash/ctid inside the transactions array cannot be resolved (and may instead try to fetch using a default/zero hash and return rpcTXN_NOT_FOUND). Refactor getTxJsonFromHistory to accept the per-transaction JSON object (or accept a parsed hash/ctid) and have getTxJsonFromParams pass txInput into it so batch lookups work correctly per element.

Copilot · 2026-03-02T22:13:40Z

src/xrpld/rpc/handlers/Simulate.cpp

+static Json::Value
+simulateTxn(
+    RPC::JsonContext& context,
+    std::vector<std::shared_ptr<Transaction>> transactions,


transactions is passed by value, causing an avoidable copy of the vector (and its shared_ptr elements) when calling simulateTxn(context, transactions, ...). Pass transactions as std::vector<std::shared_ptr<Transaction>> const& (or std::span<std::shared_ptr<Transaction> const>) to avoid copying at request scale (up to MAX_SIMULATE_TXS).

Suggested change

std::vector<std::shared_ptr<Transaction>> transactions,

std::vector<std::shared_ptr<Transaction>> const& transactions,

Copilot · 2026-03-02T22:13:41Z

src/xrpld/rpc/handlers/Simulate.cpp


-    for (auto const field : {jss::secret, jss::seed, jss::seed_hex, jss::passphrase})
+bool
+checkIsCurrentLedger(Json::Value const params)


checkIsCurrentLedger takes params by value, copying the full request object. Prefer Json::Value const& params to avoid an unnecessary copy for every simulate call.

Suggested change

checkIsCurrentLedger(Json::Value const params)

checkIsCurrentLedger(Json::Value const& params)

Copilot · 2026-03-02T22:13:41Z

src/xrpld/rpc/handlers/Simulate.cpp

+            "Must include one of 'transactions', 'tx_json', 'tx_blob', "
+            "'tx_hash', or 'ctid'.");
+    }
+    // if more than one of these fields is included, error out
+    if (numParams > 1)
+    {
+        return RPC::make_param_error(
+            "Cannot include more than one of 'transactions', 'tx_json', "
+            "'tx_blob', "
+            "'tx_hash', and 'ctid'.");


New parameter error messages use single quotes around field names, while other new messages in this handler use backticks (e.g., “Cannot use tx_hash without ...”). Consider standardizing quoting style within simulate for consistency (either backticks or quotes) so client-side matching and documentation remain uniform.

Suggested change

"Must include one of 'transactions', 'tx_json', 'tx_blob', "

"'tx_hash', or 'ctid'.");

}

// if more than one of these fields is included, error out

if (numParams > 1)

{

return RPC::make_param_error(

"Cannot include more than one of 'transactions', 'tx_json', "

"'tx_blob', "

"'tx_hash', and 'ctid'.");

"Must include one of `transactions`, `tx_json`, `tx_blob`, "

"`tx_hash`, or `ctid`.");

}

// if more than one of these fields is included, error out

if (numParams > 1)

{

return RPC::make_param_error(

"Cannot include more than one of `transactions`, `tx_json`, "

"`tx_blob`, "

"`tx_hash`, and `ctid`.");

Copilot · 2026-03-02T22:13:41Z

src/xrpld/rpc/handlers/Simulate.cpp

+            "Must include one of 'transactions', 'tx_json', 'tx_blob', "
+            "'tx_hash', or 'ctid'.");
+    }
+    // if more than one of these fields is included, error out
+    if (numParams > 1)
+    {
+        return RPC::make_param_error(
+            "Cannot include more than one of 'transactions', 'tx_json', "
+            "'tx_blob', "
+            "'tx_hash', and 'ctid'.");


New parameter error messages use single quotes around field names, while other new messages in this handler use backticks (e.g., “Cannot use tx_hash without ...”). Consider standardizing quoting style within simulate for consistency (either backticks or quotes) so client-side matching and documentation remain uniform.

Suggested change

"Must include one of 'transactions', 'tx_json', 'tx_blob', "

"'tx_hash', or 'ctid'.");

}

// if more than one of these fields is included, error out

if (numParams > 1)

{

return RPC::make_param_error(

"Cannot include more than one of 'transactions', 'tx_json', "

"'tx_blob', "

"'tx_hash', and 'ctid'.");

"Must include one of `transactions`, `tx_json`, `tx_blob`, "

"`tx_hash`, or `ctid`.");

}

// if more than one of these fields is included, error out

if (numParams > 1)

{

return RPC::make_param_error(

"Cannot include more than one of `transactions`, `tx_json`, "

"`tx_blob`, "

"`tx_hash`, and `ctid`.");

Copilot · 2026-03-02T22:13:41Z

src/xrpld/rpc/handlers/Simulate.cpp

+                auto const numParams = txInput.isMember(jss::tx_json) +
+                    txInput.isMember(jss::tx_blob) + txInput.isMember(jss::tx_hash) +
+                    txInput.isMember(jss::ctid);
+                if (numParams == 0)
+                {
+                    return RPC::make_param_error(
+                        "Must include one of 'tx_json', 'tx_blob', "
+                        "'tx_hash', or 'ctid' in each transaction.");
+                }
+                else if (numParams > 1)
+                {
+                    return RPC::make_param_error(
+                        "Cannot include more than one of 'tx_json', 'tx_blob', "
+                        "'tx_hash', and 'ctid' in each transaction.");
+                }


The batching feature explicitly allows per-item tx_hash/ctid, but the updated tests only cover batched tx_json inputs (and single-item tx_hash/ctid). Add unit tests that submit a transactions array where items use tx_hash and/or ctid, and validate the handler fetches and simulates each correctly against the specified historical ledger.

mvadari added 4 commits July 31, 2025 00:49

the naive solution seems to work

9d8b7d7

clean up

fb4b933

improve test

7e16bc5

update changelog

e377a0d

mvadari requested a review from a team July 31, 2025 13:50

mvadari added API Change Added to API Changelog API changes have been documented in API-CHANGELOG.md labels Jul 31, 2025

mvadari and others added 13 commits July 31, 2025 16:33

fix tests

f84950a

add tx_hash and ctid support

b92ca51

Merge branch 'develop' into simulate-past

85e89ed

fix build issue

7110bf1

add more tests, refactor

20cb0ce

fix tests

10a6a44

first attempt at multi transaction processing (not tested, but doesn'…

c5a31da

…t break anything)

test outline (src does not work yet)

3877263

get something working

5fc32a2

get Batch mostly working

cc7585e

clean up tests

6506a60

fix more tests

3daee17

Merge remote-tracking branch 'upstream/develop' into simulate-past

60fe9f4

mvadari changed the title ~~feat: add support for simulateing transactions in past ledgers~~ feat(rpc): add support for simulateing transactions in past ledgers Aug 4, 2025

mvadari changed the title ~~feat(rpc): add support for simulateing transactions in past ledgers~~ feat(rpc): add support for simulating transactions in past ledgers Aug 4, 2025

mvadari changed the title ~~feat(rpc): add support for simulating transactions in past ledgers~~ feat(rpc): greatly enhance simulate RPC Aug 4, 2025

mvadari and others added 5 commits August 5, 2025 11:59

add more tests

0d3e30e

Merge branch 'develop' into simulate-past

07de564

fix gcc build issue

2c94917

Merge branch 'develop' into simulate-past

c3a3ae7

Merge branch 'develop' into simulate-past

f161ab7

Copilot AI reviewed Feb 13, 2026

View reviewed changes

mvadari added 4 commits February 13, 2026 17:52

fix build issue

3f75ab0

Merge commit '25cca465538a56cce501477f9e5e2c1c7ea2d84c' into simulate…

ae947f3

…-past

Merge commit '2c1fad1023' into simulate-past

5c6ebe7

fix formatting

44a947e

Copilot AI review requested due to automatic review settings February 24, 2026 22:22

Merge branch 'develop' into simulate-past

742c282

Copilot started reviewing on behalf of mvadari February 24, 2026 22:22 View session

Copilot AI reviewed Feb 24, 2026

View reviewed changes

mvadari added 3 commits February 24, 2026 17:47

fix

406e18f

Merge branch 'develop' into simulate-past

b1f7dbb

fix build issue

876bed8

Copilot AI review requested due to automatic review settings February 26, 2026 17:15

Copilot started reviewing on behalf of mvadari February 26, 2026 17:15 View session

Copilot AI reviewed Feb 26, 2026

View reviewed changes

mvadari added 2 commits February 26, 2026 12:24

oops

8f4a532

fix import

55b8469

Copilot AI review requested due to automatic review settings February 26, 2026 19:59

Copilot started reviewing on behalf of mvadari February 26, 2026 20:00 View session

Copilot AI reviewed Feb 26, 2026

View reviewed changes

mvadari and others added 2 commits February 26, 2026 15:48

fix build

e0316e6

Merge branch 'develop' into simulate-past

c65c49a

Copilot AI review requested due to automatic review settings February 27, 2026 18:44

Copilot started reviewing on behalf of mvadari February 27, 2026 18:46 View session

Copilot AI reviewed Feb 27, 2026

View reviewed changes

mvadari added 2 commits February 27, 2026 16:41

Merge branch 'develop' into simulate-past

e689976

Merge branch 'develop' into simulate-past

2836a7d

Copilot AI review requested due to automatic review settings March 2, 2026 22:06

Copilot started reviewing on behalf of mvadari March 2, 2026 22:11 View session

Copilot AI reviewed Mar 2, 2026

View reviewed changes

+    // If this is a multi-transaction simulate request, require that each
+    // transaction explicitly specify Sequence or TicketSequence so that
+    // we do not autofill potentially conflicting values from the same
+    // base ledger view for multiple transactions.
+    bool const isMultiTxRequest =
+        context.params.isMember(jss::transactions) &&
+        context.params[jss::transactions].isArray() &&
+        context.params[jss::transactions].size() > 1;
+    if (isMultiTxRequest)
+    {
+        bool const hasTicketSeq = tx_json.isMember(sfTicketSequence.jsonName);
+        bool const hasSeq = tx_json.isMember(sfSequence.jsonName);
+        if (!hasTicketSeq && !hasSeq)
+        {
+            Json::Value jvResult = Json::objectValue;
+            jvResult[jss::error] = rpcINVALID_PARAMS;
+            jvResult[jss::error_message] =
+                "When submitting multiple transactions in a single simulate "
+                "request, each transaction must specify either Sequence or "
+                "TicketSequence.";
+            return Unexpected(jvResult);
+        }
+    }

	auto const networkId = context.registry.getNetworkIDService().getNetworkID();
	auto const networkId = context.app.getNetworkIDService().getNetworkID();

	for (int i = 0; i < batchResults->size(); ++i)
	for (std::size_t i = 0; i < batchResults->size(); ++i)

	return std::make_shared<Transaction>(stTx, reason, context.registry);
	return std::make_shared<Transaction>(stTx, reason, context.app);

	if (result.applied \|\| flags & tapDRY_RUN)
	if (result.applied \|\| (flags & tapDRY_RUN))

		to.rawTxInsert(tx.getTransactionID(), sTx, sMeta);
		apply(to);

-    context.loadType = Resource::feeMediumBurdenRPC;
+    // Set load type based on how many transactions are being simulated.
+    std::size_t numSimulatedTxs = 0;
+    if (context.params.isMember(jss::transactions) &&
+        context.params[jss::transactions].isArray())
+    {
+        numSimulatedTxs = context.params[jss::transactions].size();
+    }
+    else
+    {
+        // At most one of tx_json, tx_blob, tx_hash, or ctid is expected.
+        // If none are present, numSimulatedTxs remains 0 and the validation
+        // below will return an error.
+        numSimulatedTxs = 1;
+    }
+    context.loadType = numSimulatedTxs > 1
+        ? Resource::feeHighBurdenRPC
+        : Resource::feeMediumBurdenRPC;

	std::vector<std::shared_ptr<Transaction>> transactions,
	std::vector<std::shared_ptr<Transaction>> const& transactions,

	checkIsCurrentLedger(Json::Value const params)
	checkIsCurrentLedger(Json::Value const& params)

Conversation

mvadari commented Jul 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

High Level Overview of Change

Context of Change

Type of Change

API Impact

Before / After

Test Plan

Future Tasks

Uh oh!

codecov bot commented Jul 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Bronek commented Aug 8, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 24, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

mvadari commented Jul 31, 2025 •

edited

Loading

codecov bot commented Jul 31, 2025 •

edited

Loading