feat: implementation of MsgBuyDataAccessNFT Tx

[0xHansLee]

Implemented MsgBuyDataAccessNFT Tx, not queries.
As we discussed before, if the data pool is activated, the data access NFT is minted to the buyer. Otherwise, the buyer is added to white list of the pool. They are given the NFT when the pool is activated (maybe gyuguen is working now).

I use incrementing number as a id of NFT(token_id in contract), like stargaze. It is reasonable because each pool has each contract instance, and the id is used in each contract instance.

[0xHansLee]

I also add [ (gogoproto.nullable) = false ] to some fields.

When I export genesis.json with multiple values of those fields, all the same values (I guess the last one) are stored and exported. I referred to other projects and they all used the gogoproto extension. I'm not sure of the reason, it was a workaround for now. If you have any idea, please leave some comments.

[gyuguen]

Are you saying this?
expected:

{
  "whiteList": [
    {
      "data_pool_code_id" : 1,
      "data_pool_nft_contract_adress": "panaceaxxxx1"
    },
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    }
  ]
}

acutally

{
  "whiteList": [
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    }
  ]
}

[0xHansLee]

In your example,

{
  "whiteList": [
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    },
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    }
  ]
}

The last one was duplicated..

[gyuguen]

That's interesting...

[0xHansLee]

There are also differences of types in GenesisState. Without gogoproto extension, pointer is added to each type.

Without [(gogoproto.nullable) = false],

type GenesisState struct {
    DataValidators []*DataValidator `protobuf:"bytes,1,rep,name=data_validators,json=dataValidators,proto3" json:"data_validators"`
    NextPoolNumber uint64           `protobuf:"varint,2,opt,name=next_pool_number,json=nextPoolNumber,proto3" json:"next_pool_number,omitempty"`
    Pools          []*Pool          `protobuf:"bytes,3,rep,name=pools,proto3" json:"pools"`
    Params         *Params          `protobuf:"bytes,4,opt,name=params,proto3" json:"params"`
}

With [(gogoproto.nullable) = false],

type GenesisState struct {
    DataValidators []DataValidator `protobuf:"bytes,1,rep,name=data_validators,json=dataValidators,proto3" json:"data_validators"`
    NextPoolNumber uint64          `protobuf:"varint,2,opt,name=next_pool_number,json=nextPoolNumber,proto3" json:"next_pool_number,omitempty"`
    Pools          []Pool          `protobuf:"bytes,3,rep,name=pools,proto3" json:"pools"`
    Params         Params          `protobuf:"bytes,4,opt,name=params,proto3" json:"params"`
}

[youngjoon-lee]

There are also differences of types in GenesisState. Without gogoproto extension, pointer is added to each type.

This behavior is intended. But still, it's weird that the list contains all the same value.

[0xHansLee]

�> Is this issue still being raised? Then, I will take a look at it.

This issue is resolved now, but I don't know what's going on.

This behavior is intended.

Oh I see. I'm curious this part as well. I'll look at gogoproto extension more.

[youngjoon-lee]

I just removed [(gogoproto.nullable) = false] and checked that the exported list in the genesis contains two different elements. As I know, [(gogoproto.nullable) = false] shouldn't violate the fundamental behavior of protobuf marshalling/unmarhalling.

In protobuf, the nullability is a bit tricky to handle because everything is optional by default in protobuf 3. That's why all fields are defined with pointer such as []*DataValidator or *Params. Previously, the protobuf 2 had the required keyword which sets some field not nullable, but that feature was deleted in protobuf 3: protocolbuffers/protobuf#2497.

That's why the gogoproto, which is an extension of protobuf, introduced a new feature nullable. If you really want to make each list element not nullable, then you can use [(gogoproto.nullable) = false] so that gogoproto defines the list as []Element, not []*Element. I also agree with using it.

[youngjoon-lee]

But still, it's weird that the list contains all the same value.

I guess some Go code wasn't implemented correctly at that time.

[0xHansLee]

Oh, I see. I'm very thankful to you for your �detailed explanation.

I guess some Go code wasn't implemented correctly at that time.

I think it could be, because as I remember It was exported correctly when I implement ExportGenesis() of module params in #294.

[gyuguen]

Thank you.
I left a few comment :)

[gyuguen]

Are you saying this?
expected:

{
  "whiteList": [
    {
      "data_pool_code_id" : 1,
      "data_pool_nft_contract_adress": "panaceaxxxx1"
    },
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    }
  ]
}

acutally

{
  "whiteList": [
    {
      "data_pool_code_id" : 2,
      "data_pool_nft_contract_adress": "panaceaxxxx2"
    }
  ]
}

[inchori]

LGTM 👍

[gyuguen]

LGTM 👍

[youngjoon-lee]

At

panacea-core/x/datapool/keeper/pool.go

Line 59 in fcab598

return []types.DataValidator{}, err

, you don't needa return an empty slice which allocates memory. Just return nil, err. The zero value of slice is nil.

[youngjoon-lee]

lgtm

[inchori]

LGTM 👍

[0xHansLee]

you don't needa return an empty slice which allocates memory. Just return nil, err. The zero value of slice is nil.

Thank you @youngjoon-lee !
It is changed in 77df372

[gyuguen]

We must distribute a DataPass to a seller every time it is sold. So, when this PR is merged, I will add it to distribute to sellers whenever DataPass is sold.

[gyuguen]

LGTM 👍