Whenever I've had tools a platform/infra team provides that feels like I'm working with my hands tied, maybe it makes things withing some predicted boundaries a

Whenever I've had tools a platform/infra team provides that feels like I'm working with my hands tied, maybe it makes things withing some predicted boundaries and everything else hard, this seems to be the culprit. I never realized it until I had a teammate redesign an infra tool from a framework into a library, explicitly pointing out this distinction. Now I explicitly think about it as a fuzzy spectrum when I'm writing what's kinda analogous to higher order code, or really anything that feels like "investment" code I want to leverage to help myself down the line. Being aware of the spectrum has let me be intentional about my choices in a way that may be the biggest improvement in my programming since I learned about unit tests. Recently, pytorch-lightning has been interesting to me because it seems to be relatively close to the boundary between these two ideas.