Query node time format

[bedeho]

We want to start preserving query node state across chains, not only before mainnet, but possibly after. The problem with that is if we use block numbers, as we currently do, as the only way to time some event in our data models, then this breaks down across chains. One alternative solution is to just use block header timestamps, this could also simplify conversion to real time in some cases, in particular if block arrival times will end up being very irregular, or even explicitly changed at different points in time. The problem with them is that its hard to identify a specific block with that representation in lots of tools and apps. The most flexible solution could be to introduce some combined representation of block number, timestamp and a network identifier.

[bedeho]

Here is an example implementation

enum Network {
Babylon,
Alexandria,
Rome,
...
}

type Membership @entity {
...
 created_in_block: BigInt!,
 created_at_timestamp: BigInt!,
 created_in_nework: Network!
}

[bedeho]

Given that lots of different entities will want to re-use this three part description of a moment in time by introducing a new @field type, it would be better if we could do

enum Network {
Babylon,
Alexandria,
Rome,
...
}

type Moment @field {
 block: BigInt!,
 timestamp: BigInt!,
 nework: Network!
}

type Membership @entity {
...
 created_at: Moment!,
}

type MemberFiredEvent @entity {
...
 happened_at: Moment!,
}

[bedeho]

A last variation that, at least in this particular case can sidestep the need for introducing new @field type

enum Network {
Babylon,
Alexandria,
Rome,
...
}

type Block @entity {
 block: BigInt!,
 timestamp: BigInt!,
 nework: Network!

 /* add lots of reverselookup meta fields here for checking everything that happened in a given block */
}

type Membership @entity {
...
 created_in: Block!,
}

type MemberFiredEvent @entity {
...
 happened_in: Block!,
}