Block<T>.ProtocolVersion & BlockHeader.ProtocolVersion

[dahlia]

This implements #1142.

[codecov]

Codecov Report

Merging #1147 (849e3c0) into main (9934f3a) will decrease coverage by 0.19%.
The diff coverage is 75.47%.

@@            Coverage Diff             @@
##             main    #1147      +/-   ##
==========================================
- Coverage   87.60%   87.41%   -0.20%     
==========================================
  Files         346      347       +1     
  Lines       30049    30292     +243     
==========================================
+ Hits        26324    26479     +155     
- Misses       2092     2112      +20     
- Partials     1633     1701      +68     

Impacted Files	Coverage Δ
...net/Blocks/InvalidBlockProtocolVersionException.cs	35.71% <35.71%> (ø)
Libplanet.Tests/Common/Action/DumbAction.cs	79.29% <38.46%> (-2.87%)	⬇️
Libplanet.Tests/Blocks/BlockHeaderTest.cs	63.03% <51.75%> (-9.00%)	⬇️
Libplanet/Blocks/Block.cs	88.40% <53.33%> (-0.83%)	⬇️
Libplanet.Tests/TestUtils.cs	83.64% <77.77%> (-0.07%)	⬇️
...sts/Blockchain/BlockChainTest.ValidateNextBlock.cs	94.80% <94.73%> (-0.12%)	⬇️
Libplanet.Tests/Blocks/BlockFixture.cs	100.00% <100.00%> (ø)
Libplanet.Tests/Blocks/BlockTest.cs	98.31% <100.00%> (+0.01%)	⬆️
Libplanet/Blockchain/BlockChain.cs	90.97% <100.00%> (+0.13%)	⬆️
Libplanet/Blocks/BlockHeader.cs	95.85% <100.00%> (+2.06%)	⬆️
... and 6 more

[dahlia]

@planetarium/libplanet PTAL.

[limebell]

Something is missing I guess...

[dahlia]

Completed the sentence.

[dahlia]

@planetarium/libplanet Rebased. PTAL.

[junorouse]

I understand that libplanet separately control the flow of actions through protocol version management (this patch) instead of creating a new action file (i.e., Buy {2,3,4}.cs), am I right? -- for proper patching of planetarium/lib9c#248

If so, this PR looks OK for me. However, from an attacker's point of view, if an attacker has enough mining power, the block's previous version can continue to be mined even if the new protocol version introduces.

There are some mitigations for it:

1. Set the maximum number of blocks(=limit) allowed with the previous protocol version.
2. Like lib9c, use an authorized miner to introduce a new version.
3. Change to a consensus algorithm to that selecting the next block proposer randomly.
4. ...

[dahlia]

@junorouse Thanks for your review and suggestions!

However, from an attacker's point of view, if an attacker has enough mining power, the block's previous version can continue to be mined even if the new protocol version introduces.

Even if we don't merge this PR, that attackers can outpace honest nodes is the critical problem in itself and we need to address it as soon as possible. Therefore, IMHO we'd better deal with it in the separate issue/patches.