Howdie, I find the rule graph thingy extraordinary useful and powerful. Would love to be able to use it in other projects. How generic is it? Thinking, would it be feasible to extract it as a separate library package for generic use.. ?

glad to hear it, although there are definitely some polish issues we’d like to resolve (…as you mention in your other thread!)

What are examples of how you could use it outside of Pants? Could you simply use Pants in those cases?

I haven’t worked out a concrete use case as example, just my head seeing possibilities. As I tend to write quite a bit of “data driven pipelines”.. that is, generic pieces that does stuff depending on input data. As a “delivery engineer” working on our in-house tooling for CI/CD and more (for traceability etc) So from that perspective, hmm.. yeah, perhaps I could leverage pants directly to do completely other stuff, using plugins etc to drive the logic and feed it generated “BUILD” files (as they wouldn’t be for building stuff..) That would most likely be easier and more reliable than attempting to rip the rule engine out of its current context into something else.. thanks for that nudge 😉

Mm, yea. There should be a fair amount that is possible to do directly in Pants. If you’re able to divide the “referentially transparent”/idempotent stuff into

@rules

and the side effecting stuff into

@goal_rules

at the top of the graph, lots of CI/CD stuff should be possible