v0.15.9

Added

• Improved WebSocket RPC methods by adding access to the pre-latest blockchain state. This will improve UX for users of subscription based methods since it will be able for them to return updated information faster.

Fixed:

• Fixed an issue with starknet_estimateFee endpoint with incorrect handling of estimations that runs out of gas.
• Fixed an issue with starknet_simulateTransactions where DEPLOY_ACCOUNT transactions could fail unexpectedly with insufficient funds error.
• Fixed an issue where pre-latest blocks didn't have a block number assigned.

Changed:

• Internal state refactors improving code quality and minor benefits to performance.

Full Changelog: v0.15.8...v0.15.9

v0.15.8

Added

1. We extended the support of the pre_latest block for the following RPC methods:

• starknet_getTransactionStatus
• starknet_getMessagesStatus
• starknet_getTransactionByHash
• starknet_getTransactionReceipt
• Support for WebSocket methods in the next release!

2. Added new CLI flag rpc-call-max-gas that sets max gas for starknet_call RPC calls. The default is the same value as the constant used in previous versions: 100 million.

Fixed:

1. Important security issue.

Full Changelog: v0.15.7...v0.15.8

v0.15.7

Added

• Support for custom chains. Eth and Strk tokens are no longer fixed and its addresses are requested to the Sequencer instead. For network such as Mainnet and Sepolia, defaults are used.

Fixed

• Bug in the latest release, which didn't give the entire benefit of improved UX

Full Changelog: v0.15.6...v0.15.7

v0.15.6

What's new

This version adds support for the pre-latest block concept — a block that has been finalized by consensus but whose commitments are not yet calculated. It acts as a preview of the next block, allowing faster access to the chain state to build the pre_confirmed state.

When the latest finalized block is N, the pre_confirmed block can now be built either:

• directly on top of N (i.e., pre_confirmed = N + 1), or
• on top of a PreLatest block at N + 1 (i.e., pre_confirmed = N + 2), if available.

Background

Previously, the pre_confirmed block was always fetched for latest + 1, requiring us to wait for commitments before proceeding.

Impact

This change allows us to fetch the pre_confirmed block earlier by removing the wait for commitments, resulting in access to a more recent state sooner and reducing transaction confirmation latency.

Full Changelog: v0.15.5...v0.15.6

v0.15.5

Changed

Improved Re-Org mechanism

Re-org mechanism was improved to trigger immediately after detecting that the Sequencer's latest block is on a lower height than the latest block the node previously received. Additionally, some verifications were included to avoid re-org'in with false positives.

This is better than the previous behaviour, where, even after detecting the reorg, we would wait for the new chain to reach the height of the previous chain, causing Juno to look stuck.

Full Changelog: v0.15.4...v0.15.5

v0.15.4

⚠️ All the node users should update to this version as soon as possible

What's Changed

• Blocks between 1 943 705 and 1 952 704 will no longer be re-executed locally in the node, but by the sequencer. Every RPC execution request (versions 8 and 9) in this block range will be forwarded to the Starknet Sequencers.

Note: user interaction with the node remains the same, as it was a normal RPC request.

Full Changelog: v0.15.3...v0.15.4

v0.15.3

What's Changed

• Added a reorg notification to receipt and transaction subscribers 3071.
• Improved support for ACCEPTED_ON_L1status on starknet_subscribeEvents endpoint in some cases 3072.
• Added an option to disable batched requests to the RPC servers 3086.
• Update blockifier dependency to the latest 0.14.0-rc.

Full Changelog: v0.15.2...v0.15.3

v0.15.2

What's new

This release comes with full support of Startnet RPC v0.9.0.

Changes

• When executing a transaction without validation there is no longer a strict nonce check. This means that an account can invoke a transaction with any nonce bigger than their current one instead of the next immediate nonce 077c3b1
• Rewrote RPC Transaction cache reducing greatly the amount of memory allocated (details ) e17670d
• Improved L1 data monitoring 62b3363
• Added docs explaining Juno's Kubernetes health endpoints 256c9c4
• Provided grafana dashboard updated to work with standard Prometheus configurations 4003007

Full Changelog: v0.15.1...v0.15.2

v0.15.1

This release improves the stability of the client

Added

• Support to RPC specs 0.9.0-rc2. Blocks can be queried by the l1_accepted block tag to get the latest Starknet block which is part of the finalized Ethereum state.
• Dynamic timeout doesn't increase on bad request error response from feeder gateway.

Fixed

• Increased constraints on when to warn when dynamic timeout increases beyond certain threshold
• Fixed an edge case in TransactionByBlockIDAndIndex RPC endpoint.
• Pending block now gets assigned latest block num + 1. Before it was 0, causing transaction that depend on it this info fail when estimating fee.

Full Changelog: v0.15.0...v0.15.1

v0.15.0

This release brings support for Starknet v0.14.0 and implements the new RPC v0.9.0.

⚠️ Important Upgrade Notice ⚠️
Nodes running older versions will stop syncing once the Starknet testnet upgrades to v0.14.0. You must upgrade to stay compatible.

Added

• Support for Starknet v0.14.0
• Implementation of RPC v0.9-rc.1

Full Changelog: v0.14.7...v0.15.0