Main net is running RNode version 0.12.2. This version includes the block merge code that does not require a hard fork, but block merge is currently disabled until we have the block merge version coming up in hard fork 2, aka block merge hard fork. RNode version 0.12.2 includes Will's bug fix for transaction API. Main net observers will be shortly updated to 0.12.3 that has Will's PR to alleviate the tuplespace mismatch error.
There two planned hard forks: one for block merge (release versioning will be 0.13.x) and one for PoS changes (release versioning will be either 0.14.x or 1.x.x) to eventually enable a fully decentralized node with external validators.
Both *.bm.testnet.rchain.coop and *.llbm.testnet.rchain.coop are part of hard fork 2. the llbm version has the latest code. Having the two testnets will allow the community testers to trace any unexpected behavior in their tests to the latest code changes vs. the code that has been running on bm.testnet for a while. Current state of the llbm testnet is that it requires network restart every couple of days or so. The team is resolving any bugs/ reliability issues on *.llbm.testnet.rchain.coop (Leaderless blockmerge testnet) . However, we are making it available for community testing, so we get early feedback. Community members interested in testing can request testrev for llbm in the #testnet channel. Once we determine llbm.testnet to be stable, it will become the sole testnet for hard fork 2.
Sprint 100 in progress
Main Focus is to prepare the block merge Hard fork (hf2), specifically (a) merging REV balances directly - currently on hold until other llbm issues are resolved (b) leaderless block merge - work ongoing (c) a lot of testing + bug fixing and (d) performance improvements. Other work includes resolving any identified bugs, hardening the network. Current PR list is at https://github.com/rchain/rchain/pulls .
Block merge Hard Fork (hf2)
The team is reviewing the recent designs for llbm (see item 7 below), clarifying and prioritizing the work that needs to be done to complete HF2 and the distribution of those tasks among the team members. The main focus is to discuss and complete the design changes, stabilize the code, rewrite tests, make sure everything works as expected.
Will has added debugging code to get more information about the source of the 'tuplespace mismatch' error and running this code on a separate machine, waiting for the error to occur, so that we can capture information on what's causing the error.
Denis is making progress on his algorithm for radixTree implementation of History to increase concurrency and reduce the amount of data that needs to be handled and stored.
Gurinder has been working on testing the llbm network, capturing the state information when errors occur. He's also trying to replicate tuplespace mismatch error.
At this time, the llbm testnet is still unstable, but the team is fixing the problems one by one. More details in item 6 below and nutzipper's presentation.
From nutzipper - Code restructuring changes discovered as part of the LLBM work (hopefully this helps the community to understand why these are substantial changes and take time to implement, test and correct):
Current view of the node only holds blocks that are validated. This is wrong concept - we have to have a view on the most recent state of the network, to be able to see the latest finalized state also
Replaying of the block is not necessary for maintaining the most recent view - justification structure is what defining it. So only observing of signed messages is enough.
Validation is separate from observation, it starts from the moment when finalization fringe is found, then validator can replay blocks, building the state of the network.
Validation of the state is only required if you want yourself to be sure that the network is making the right decision, or if you want to propose changes to the state (you are bonded validator). Otherwise, just running lightweight Casper client is enough to follow finalization and find out when your deploy is finalized.
Validation of the full state is required only to propose changes, the target is to eventually be able to validate only part of the state (e.g. rev vault balances).
Validation part has to be adjusted / simplified. Equivocation detection is now not part of validation of a particular message, but part of block creation process, when the latest network scope is defined.
We're moving towards keeping state of the Casper node in a single class, modified atomically, to make sure we eliminate all kids of races.
We're reorganizing the storage to have efficient access to data, e.g. to not load the whole block to just get access to some fields.
Release Plan for the rest of the year and beyond - for reference
Currently planned near term releases are
Block merge release hard fork 2 plus any configuration changes that do not need extensive work - Leaderful and Leaderless block merge
Configuration changes release,
Performance improvements releases (one or more),
Hard Fork 3 with PoS changes and other items,
External validators process validation with trusted community validators (aka Friends & Family) on the test net
Friends & Family validation on the main net
External validators among extended ecosystem (e.g. exchanges, if they're willing)
External validators in the wild.
Explanation of the different Hard Forks (for reference)
Proceeding further with block merge requires changes to the per validator vault to make it a per deploy vault/purse etc. We are currently finalizing the design for this. Actual code and test will take some time. We are potentially looking at this scenario: 1. First hard fork for balances and removing the slashed validator 2. Second hard fork for block merge - This is the fastest way to get block merge on the main net without holding it until the third hard fork. This includes REV balance merging (aka per deploy vault), attestation messages and the Tree depth being increased to 4. 3. Third hard fork with all other changes. Third hard fork includes implementation of a lot of changes including the PoS contract changes, Soft fork process etc. Design for soft fork process is complete and is being reviewed by the team.
(Hard Fork) TEST NET 2 ADVISORY: (for reference)
To test the version of block merge with data changes that need a hard fork, we are creating a separate feature branch and a TEST NET 2 as the testbed for this feature branch. This content will be left here until we are close to Hard Fork 2 implementation on the main net. TestNet 2 is now available for the community to test their contracts on.
There will be no guarantees of data storage format compatibility on Test Net 2, as the data storage format changes are incrementally implemented. The anticipated changes are mostly storage level format changes visible only to node operators. We encourage the community to start using Test Net 2 in addition to the current test net, so that we can quickly identify and resolve any issues with the upcoming hard forks. All new development should be targeted to TEST NET 2 and all current code MUST BE tested against TEST NET 2 to ensure future compatibility.
On all the hard forks on the main net, we will be starting from an empty state, with REV balances only. On Test Net 2, this 'loosing the state' will be repeated multiple times.
Tech-Governance meetings on Fridays 9:30 AM Eastern, 6:30 Am Pacific