IF: Add deep-mind ACCEPTED_BLOCK_V2

[heifner]

Add deep-mind ACCEPTED_BLOCK_V2 which will be exported for blocks after transition to Savanna.

ACCEPTED_BLOCK_V2

• [block_num] [last-irreversible-block-num] [signed_block_packed_in_hex] [finality_data_packed_in_hex]

struct finality_data_t {
   uint32_t     major_version{light_header_protocol_version_major};
   uint32_t     minor_version{light_header_protocol_version_minor};
   uint32_t     active_finalizer_policy_generation{0};
   digest_type  action_mroot{};
   digest_type  base_digest{};
};

Changed DEEP_MIND_VERSION to spring 1 0.

Resolves #16

[linh2931]

No need to save the deep mind log file for tests? You might add a test for this new version or create an issue for it.

[heifner]

No need to save the deep mind log file for tests? You might add a test for this new version or create an issue for it.

Could add transition to the deep mind test. But I wonder if there is any point. That test is a pain to maintain and not sure it would add any benefit to add a much larger amount of logs to update when it changes. I can add it if we think it adds value.

[linh2931]

No need to save the deep mind log file for tests? You might add a test for this new version or create an issue for it.

Could add transition to the deep mind test. But I wonder if there is any point. That test is a pain to maintain and not sure it would add any benefit to add a much larger amount of logs to update when it changes. I can add it if we think it adds value.

Agreed. No need to add transition to the test. Just a test of v2 in Savanna mode.

[heifner]

Just a test of v2 in Savanna mode.

Not sure what you mean by this. My point is that the only deep-mind test we have is difficult to maintain and provides little to any benefit in my opinion. I would hate to add another that also requires blindly updating anytime we touch something that changes the format of traces.

[linh2931]

Just a test of v2 in Savanna mode.

Not sure what you mean by this. My point is that the only deep-mind test we have is difficult to maintain and provides little to any benefit in my opinion. I would hate to add another that also requires blindly updating anytime we touch something that changes the format of traces.

I think the value of the existing deepmind test is not the format checking but to make sure basic deepmind logging is still working after some big changes in other parts of the code base. No need to add a new one if too much effort is required for creating and maintaining.

[greg7mdp]

Why not have two lambdas instead of this if constexpr? The if (auto* dm_logger = get_deep_mind_logger(false)) { could be before the apply call.

[spoonincode]

You might be able to use a single "overloaded lambda" here? I settled on that being the most attractive pattern personally, for example,
https://github.com/AntelopeIO/leap/blob/0dd9c26896ec27850bafadbd65839bbd41455d35/libraries/chain/controller.cpp#L1068

[ericpassmore]

Note:start
group: IF
category: INTERNALS
summary: Add deep-mind ACCEPTED_BLOCK_V2 which will be exported for blocks after transition to Savanna.
Note:end