You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the future, the rusk binary should include all features by default. Then it should be decided during runtime which features will be enabled and ran or not.
Detailed Description
The goal is to provide one rusk binary that can act as the different node types, rather than providing several different binaries. The decision which feature is enabled is then made by a CLI flag and possibly an additional settings file.
In addition to that, we want to keep the cfg flags approach to retain the functionality of providing nodes that do not contain the bytecode of features that are not used.
Currently there can be 3 different types of nodes:
Full node (Provisioner)
Full node + prover (prover feature enabled)
Archive node
Possible Solution
In general, it might be a decent approach to have a builder pattern in the code that is able to create the different types of nodes with the specific features in it.
However, there should be some planning before this is tackled. I expect this work will overlap with other issues and create some new issues and it should be aligned with them beforehand. I also do think it will touch parts of the codebase that should be reconsidered in terms of modularization and encapsulation, especially also reviewing how specific parts of the codebase relate to each other.
The text was updated successfully, but these errors were encountered:
Summary
In the future, the rusk binary should include all features by default. Then it should be decided during runtime which features will be enabled and ran or not.
Detailed Description
The goal is to provide one rusk binary that can act as the different node types, rather than providing several different binaries. The decision which feature is enabled is then made by a CLI flag and possibly an additional settings file.
In addition to that, we want to keep the cfg flags approach to retain the functionality of providing nodes that do not contain the bytecode of features that are not used.
Currently there can be 3 different types of nodes:
Possible Solution
In general, it might be a decent approach to have a builder pattern in the code that is able to create the different types of nodes with the specific features in it.
However, there should be some planning before this is tackled. I expect this work will overlap with other issues and create some new issues and it should be aligned with them beforehand. I also do think it will touch parts of the codebase that should be reconsidered in terms of modularization and encapsulation, especially also reviewing how specific parts of the codebase relate to each other.
The text was updated successfully, but these errors were encountered: