Skip to content
Not a network amendmentTier BUpcoming standardsSpec: Draft

WASMVM XLS-102

The draft defines the deterministic WebAssembly runtime, gas rules, and host functions for XRPL extensions. This is a proposal, not an active network feature. Whether any part of it will launch, and when, remains open.

Sources reviewed Oct 5, 2026 · English page updated Oct 5, 2026 · Phil · CryptoTuts founder

Network status
Not a network amendment
XLS
XLS-102
Spec status
Draft
Spec revision
Sep 3, 2026

What does this XLS propose?

What problem is the proposal intended to solve?

WASM leaves implementation choices for the runtime, gas metering, and host access. Without a specified engine and stable ABI, nodes could evaluate the same bytecode differently and lose consensus.

How is the mechanism designed in the current draft?

Each execution receives a fresh, isolated WASM instance and a fixed gas budget. Instructions, memory operations, and host functions have deterministic costs. Only defined host functions may read ledger data or modify the host object's Data field. Nondeterministic features such as native floating-point operations, randomness, system time, and host I/O remain excluded.

How settled is the proposal?

This is a proposal, not an active network feature. Whether any part of it will launch, and when, remains open.

What changes compared with the predecessor?

BeforeAfter
The XRPL has no general consensus-compatible WASM runtime with a binding ABI and gas metering.Smart Escrows and later extensions could execute reviewed bytecode in a bounded, reproducible sandbox.

Who is this change relevant to?

XRP holders and users

XLS-102 alone gives users no new transaction. They would be affected indirectly if Smart Escrows or contracts execute code in this VM.

Wallets and app developers

Developers must work with a stable host ABI, their own memory management, and explicit buffer sizes. Ordinary system functions are unavailable in the sandbox.

Exchanges and issuers

Institutions could use deterministic compliance or release logic but must first review and simulate bytecode and gas requirements.

Node operators

Validators bear the execution costs. The UNL can adjust bytecode size, the compute limit, and gas price, or set the compute limit to zero when problems occur.

What limitations and risks remain?

  • XLS-102 is a draft and itself introduces no transaction, ledger object, or RPC.
  • The initial host-function scope applies only to Smart Escrow.
  • Each host-function call may transfer at most 1 MiB across the WASM boundary and read or write at most 1 KiB of ledger data.
  • WASM 1.0 has no automatic garbage collection. Calling code must allocate and reuse memory itself.
  • Code cannot read historical ledger data, traverse directories arbitrarily, or directly modify raw ledger objects.
  • Native WASM floating-point instructions, randomness, host-system I/O, and other nondeterministic features are excluded. Deterministic decimal arithmetic is available through the defined float host functions.
  • Deployed host functions must remain permanently backward compatible. New functions may be added, but incompatible changes require a differently named function; deprecated functions remain callable.

What fee and performance figures does the specification provide?

  • Gas is consumed by WASM instructions, memory operations, and host functions. Exceeding the budget stops execution immediately.
  • Bytecode size, the compute limit, and gas price are intended to be subject to UNL voting. XLS-102 provides no final global initial values.
  • Individual host functions have fixed draft values, such as 60 gas for ldgr_index, 5,000 gas for cache_le, and 1,000 gas for set_data.
Technical details: transactions, fields, codes and invariants

Transactions

  • no new transaction types; initial execution through EscrowFinish

Ledger objects

  • Escrow as the first host object
  • FeeSettings for limits subject to voting

Fields and flags

  • Gas / GasLimit / GasPrice
  • BytecodeSizeLimit
  • linear WASM memory pages of 64 KiB
  • set_data as the only host function that writes ledger data in the initial scope

API methods

  • ldgr_index / parent_ldgr_time / parent_ldgr_hash
  • home_le_field / home_le_inner
  • cache_le / le_field / le_inner
  • transaction, keylet, NFT, and trace host functions

Invariants

  • The same inputs must produce the same gas consumption and result on every node.
  • Each execution starts in a new VM instance without reusing memory state.

How does this proposal relate to others?

Every direct relationship has a type, an explanation and a primary source.

What happens next for this proposal?

This is a proposal, not an active network feature. Whether any part of it will launch, and when, remains open.

The draft needs further technical development and may still change fundamentally or be abandoned. Only a corresponding xrpld (formerly rippled) amendment would trigger a separate network vote.

Where does it sit within its amendment family?

WASMVM has no amendment family with other members on this site, so there is no progression to show here.

Which primary sources support this information?

← All XRPL amendments and XLS specifications