Support pickle and deepcopy on Triangulation#11
Merged
Conversation
adaptive's LearnerND does deepcopy(self.__dict__) in _get_data, which backs learner.save(), copy.deepcopy(learner), and pickling learners for distributed runners - all of which crashed with 'cannot pickle builtins.Triangulation' under the Rust backend (the reference is a plain Python class, so it pickles for free). - the constructor gains an optional simplices argument that restores an exact prior state via from_simplices instead of triangulating (no scipy, no re-insertion; hand-modified simplex sets and unconnected vertices survive) - __reduce__ serializes (vertices, simplices) and reconstructs through that argument; this covers pickle, copy.copy, and copy.deepcopy - the pyclass declares module = 'adaptive_triangulation._rust' so pickle can resolve the class by reference (it previously advertised itself as builtins.Triangulation) Tested: pickle/deepcopy round-trips (including a hand-deleted simplex and an unconnected duplicate vertex), clone independence, and an end-to-end LearnerND _get_data/_set_data + deepcopy cycle that keeps learning afterwards.
- return Bound values directly instead of unbind/into_any noise, with a PickleReduction type alias for the pickle protocol shape - sort the simplices before serializing so equal triangulations pickle to identical bytes (the hash-map iteration order is arbitrary)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Found via LearnerND integration:
LearnerND._get_data()doesdeepcopy(self.__dict__)— which backslearner.save(),copy.deepcopy(learner), and pickling learners for distributed runners — and all of them crashed withTypeError: cannot pickle 'builtins.Triangulation'under the Rust backend. The Python reference is a plain class, so it pickles for free; this was a real drop-in gap (in v0.1.0 and v0.2.0).Three pieces:
simplicesargument that restores an exact prior state viafrom_simplicesinstead of triangulating — no scipy, no re-insertion, so hand-modified simplex sets and unconnected vertices survive the round-trip__reduce__serializes(vertices, simplices)and reconstructs through that argument; one implementation coverspickle,copy.copy, andcopy.deepcopymodule = "adaptive_triangulation._rust"so pickle can resolve the class by reference (it previously advertised itself asbuiltins.Triangulation)Tests
vertex_to_simplices, andreference_invariant()— including a hand-deleted simplex (exact-state restore, not re-triangulation) and an unconnected duplicate vertexcopy.deepcopy(learner)and a_get_data()/_set_data()cycle, with both restored learners continuing to learn afterwards115 pytest + 18 cargo tests pass; pre-commit clean. Suggest tagging v0.2.1 once merged.