Hi everyone, Regarding the `arrow.big_decimal` extension proposal, we initially started this effort to introduce a canonical type capable of representing arbitrary precision and scale. However, based on the feedback received so far, it's clear that standardizing this across Arrow requires a broader scope of work, especially around compute kernel support and alignment with how different tools in the ecosystem handle arbitrary-precision numerics.
We think it makes sense to pause active work on the current proposal for now. Rather than pushing forward as-is, we want to take a step back to gather more information and look more closely at the various decimal types and patterns currently used across the community. Once we have a clearer picture of existing ecosystem practices and specific user requirements, we can re-evaluate how best to approach this. In the meantime, we’d love to hear your thoughts, use cases, or details on how your projects currently handle arbitrary-precision decimals! Finally, a huge thanks to Micah, David, Matt, Curt, and others for their valuable insights and contributions to the discussion so far. Best, Mandy
