CLN: MultiIndex and Index no longer inherit from ndarray. · Issue #5080 · pandas-dev/pandas (original) (raw)

Skip to content

Navigation Menu

• • GitHub Copilot Write better code with AI
  • GitHub Advanced Security Find and fix vulnerabilities
  • Actions Automate any workflow
  • Codespaces Instant dev environments
  • Issues Plan and track work
  • Code Review Manage code changes
  • Discussions Collaborate outside of code
  • Code Search Find more, search less
• Explore
  • Learning Pathways
  • Events & Webinars
  • Ebooks & Whitepapers
  • Customer Stories
  • Partners
  • Executive Insights
• • GitHub Sponsors Fund open source developers
  • The ReadME Project GitHub community articles
• • Enterprise platform AI-powered developer platform
• Pricing

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Sign up

Appearance settings

Description

Changing to not be an ndarray subclass should make a number of things simpler (especially the API unification discussed in #3268). Plus, it will clarify the required interface for an Index object. It avoids all the workarounds to make Index immutable (instead, will define __array__ and __array_wrap__ and choose which other functions should be allowed).

• [ ] Move Index underlying data to _data to mirror other objects. Keep metadata on object itself. (MI should work with only minor tweaks, given that it was always a fake ndarray)
• [ ] Reimplement fastpath. (figure out how to do this inference again).
• [ ] Move MI underlying data to _data (potentially turn levels into 2d int64 ndarray and keep labels as list of lists or as ordered-dict-like. not sure whether row-major or column-major makes more sense. I'm guessing row-major since that would speed key lookups)

How much backwards compatibility do we need to support? (pickle compat should be simple). E.g. do we want to do something like this:

def getattr(self, attr): res = getattr(self._data, attr) warnings.warn("Accessing %s on %s is deprecated. Convert to an ndarray first." %
(attr, self.class.name)) return res