Selective Block Execution Using Cached Outputs for Efficient Pipeline Debugging

[zah-tane]

Description of the requested feature:

It would be more efficient if there were a way to select specific blocks within a pipeline and, instead of running them every time the entire pipeline is executed, use the previous outputs of those blocks in the current run. This feature would be helpful especially when debugging a block in a complex pipeline that appears later in the pipeline graph.

Brief explanation on why you want/need said feature:

It saves money to reuse cached block outputs for those blocks that call an API and also saves time when debugging pipelines.

Do you intend on coding this feature yourself?

No

Which zetaforge components would be affected by this new feature:

• client
• anvil
• driver

Share any code snippets related to the feature here(optional):

No response

Provide an API design for the feature(optional):

No response

[FGRCL]

Maybe a combination of the block's input and its tag would be sufficient to determine if a block needs to be rerun