TEST.3.0

• heavy groundwork on deploying a new client and JSON-RPC API
• massive performance improvements after profiling through algorithm updates, caching, indexing and pre-hashing
• massively reduced redundant inter-node communication through earlier and more precise caching, and reducing the max number of incoming connections to 20
• various documentation improvements
• added EditorConfig file
• various logging improvements (removed spurious logs, added miss and implicit roll sell warns)
• improved client user interaction
• various code clarity improvements
• test suite improvements
• added ouf-of-the-box IPV6 compatibility
• corrected a timing bug causing slots to be slightly shifted in time
• improved the bootstrap process (breaking)

Full Changelog: TEST.2.2...TEST.3.0

TEST.2.2

• update various libraries to the latest version to avoid some compilation problems with the latest rust nightly

Full Changelog: TEST.2.1...TEST.2.2

TEST.2.1

• Added a caching system to prevent over-propagation of headers and endorsements
• disabled TCP_NODELAY to increase performance at the expense of a slightly higher latency
• Added tests
• Improved docs

Full Changelog: TEST.2.0...TEST.2.1

TEST.2.0

• main change: endorsement system, massively increasing architecture-based attack costs and making attack attempts fully prohibitive (see scientific paper)
• changed the block reward to 0.3 coins
• changed roll price to 100 coins per roll
• changed the periods_per_cycle parameter from the temporary debugging "10" value to the final "128" value, thus making cycles much longer and reducing the probability that a single miss causes an implicit roll sale
• config override: now, the massa-node/base_config (and massa-client/base_config) folder contains configuration files that shouldn't be edited by the user while massa-node/config (and massa-client/config) contain user-editable files that override the base config and also contain the node_privkey and staking_keys files. This makes massa docker-friendly.
• the massa-node/storage folder now contains all the storage data, in order to be container-friendly and pave the way to docker CD
• added warnings on misses and implicit roll sales happening on addresses staked by the node
• internal rework of the block_graph system to allow retrieving clique fitnesses after computation, and always knowing which clique is the blockclique
• various improvements on logging, UX, docs, performance and stability

Full Changelog: TEST.1.3...TEST.2.0

TEST.1.3

• improved logs
• improved docs
• groundwork for the implementation of endorsements
• improved desync detection security
• improved consensus production stats flexibility and accuracy
• client: moved staker stats from staker_info to a batched version named staker_stats
• improved code clarity and stability

Full Changelog: TEST.1.2...TEST.1.3

TEST.1.2

• improved log clarity
• improved config defaults for security
• improved documentation
• limit the number of attempts at filling a block with operations
• improved the next_draws command
• increased stats duration to 3 cycles for deactivation security
• added block_ids_by_creator to list blocks created by an address
• added warning in client when performing operations that require more rolls or coins than what the address has
• node banning was re-enabled with a 1 hour unban timeout

Full Changelog: TEST.1.1...TEST.1.2

TEST.1.1

• improved documentation
• reduced functional test boilerplate thanks to a wrapper function
• clarified logs, and demoted logs that disturbed people
• clarified the results displayed by some client commands
• added the possibility of adding an existing private key to the wallet
• removed all compilation warnings
• bootstrap servers are now tried in random order to avoid overloading the 1st one
• prevent operations from being propagated to nodes that are aware of it (according to the local node's vision)
• operations inside blocks are now added to the operation pool
• the node_sign_message endpoint was added to the API and allows signing arbitrary messages using the node's private key
• the node's ID (public key) is now displayed in the logs on startup, but also available in the network_info client command (and matching API command)
• added command-specific help in the client. Now you can type "help network_info" to get the help on network_info
• now, the list of nodes you are actively connected to is available in the "peers" command, together with the remote node ID (public key) and connection direction (in or out)
• a new command "testnet_rewards_program" was added to the client and allows you to prove you're the owner of a node ID, a staking address, and a discord identifier
• the staking_keys file was removed from git to avoid overwriting yours
• when launching the client with "cargo run --release" (without specifying a wallet), "wallet.dat" will be used by default, making all commands available
• all currency values are now wrapped into an "Amount" structure, ensuring safe decimal arithmetic and consistent display of amounts of coins
• added the config parameter network.target_bootstrap_connections limiting the number of connections from/to bootstrap nodes
• added the config parameter network.max_out_bootstrap_connection_attempts to limit the number of simultaneous outgoing connection attempts towards bootstrap nodes
• added warnings telling the user to avoid sharing their private keys every time they are displayed
• added the end_timestamp config parameter to self-destruct expired nodes when an episode ends
• stakers that miss a high percentage of their block productions during cycle C will cause an implicit roll_sell of all their rolls at the beginning of cycle C+3
• a versioning system was added to avoid bootstrapping/connecting to nodes running incompatible versions of the protocol
• per-cycle final block production statistics and miss rates were added to staker info
• node desync is now detected in about 1 minute, and auto-resync is attempted
• adjusted various config values
• simplified bootstrap initiation for easier deployment
• various bug and usability fixes + optimizations

Full Changelog: TEST.1.0...TEST.1.1

TEST.1.0

False start. Do not use this version.

Full Changelog: https://github.com/massalabs/massa/commits/TEST.1.0