Skip to content

Goal: Duplicate record pruning #1414

Description

@vinarmani

Issue raised by trufnetwork/adapters#237

There are many streams on the network that consist of many duplicate entries. In some cases, the value of a stream has never changed from the first data point broadcast, yet there is a data point, with the same exact value, that was broadcast every 24 hours. There is no application where such repeat data points are necessary on the network.

  • In the case of a query for "current value," the current value is always the last point broadcast. For instance, although a market may not be open on a weekend, during the time that the market is closed, the last value broadcast (the closing price on Friday) is considered the "current price."
  • In the case of certified attestation or prediction protocol settlement, the attested value is the most recent value at the time of the request/settlement. It doesn't matter if that value was broadcast 1 month ago or one minute ago. Therefore, if the value has not changed in a month, the attestation of the month-old value is as valid as the attestation of the one-minute-old value.

In the same way that pruning of high-frequency streams is done, leaving only open-high-close values, pruning of repeat values older than a certain threshold (ie. 30 days), would provide significant reductions to resource requirements (storage) for network participants over time and at scale.

Deadline

ETA: undefined

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions