Routing Phase 3: export the recorded problem to a host representation#1583
Routing Phase 3: export the recorded problem to a host representation#1583ramakrishnap-nv wants to merge 3 commits into
Conversation
Add to_host_problem(dm): walks the store-then-build IR (DataModel._calls) and produces host (numpy) arrays keyed by the gRPC RoutingProblem field names, exporting any device (cuDF/cupy) inputs to host at this point (the mixed-IR "export on serialize"). Proto-agnostic -- a dict of host arrays -- so it does not depend on generated gRPC stubs and is the client-side foundation for serializing a routing problem over gRPC. Covers the full setter surface, including uniform and per-vehicle break dimensions. Verified end-to-end: host+device (numpy/cuDF) inputs export to a fully host problem with correct row-major matrices, grouped per-vehicle breaks, and no device references left; full routing suite passes (56). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: Ramakrishna Prabhu <ramakrishnap@nvidia.com>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
Replace the per-setter if/elif chain in to_host_problem with a dispatch table: setters that rename fields, take several args, or build nested structures get an explicit handler in _HANDLERS; the common set_<field>(array) -> field <field> setters are derived automatically and need no entry. An unmapped add_* setter raises, and test_every_setter_is_exportable checks every recorded setter is exportable, so a new setter fails loudly instead of being silently dropped. Behavior is unchanged. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: Ramakrishna Prabhu <ramakrishnap@nvidia.com>
|
@CodeRabbit review |
✅ Action performedReview finished.
|
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Enterprise Run ID: 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (2)
📝 WalkthroughWalkthroughChangesRouting serialization
Estimated code review effort: 3 (Moderate) | ~20 minutes 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@python/cuopt/cuopt/routing/_serialize.py`:
- Around line 180-184: Update the public function to_host_problem with explicit
parameter and return type annotations, using the project’s existing types for
the routing problem input and host-array mapping. Expand its docstring to
document the dm parameter, returned RoutingProblem-keyed host arrays, and the
KeyError raised when required problem fields are missing.
- Around line 194-205: Update the unregistered set_* fallback in the _serialize
loop to require exactly one argument before exporting args[0]; raise the
existing unmapped-setter error (or equivalent) for multi-argument calls so they
require an explicit _HANDLERS entry. Add a regression test covering a
multi-argument set_* call and verify it is rejected.
In `@python/cuopt/cuopt/tests/routing/test_serialize.py`:
- Around line 61-77: Strengthen the routing serialization assertions in
python/cuopt/cuopt/tests/routing/test_serialize.py lines 61-77 by validating
time-window arrays, both capacity demand/capacity arrays, and matrix dtype plus
metadata alongside values, shapes, and existing fields. Also update lines 94-103
to assert uniform-break earliest/latest and every field of both vehicle breaks,
including both locations arrays; retain the existing device-reference check.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 924e8e2a-f40f-4a0a-9a2c-9a10478710cf
📒 Files selected for processing (2)
python/cuopt/cuopt/routing/_serialize.pypython/cuopt/cuopt/tests/routing/test_serialize.py
- to_host_problem: add parameter/return type hints and a full docstring (parameters, returns, KeyError behavior). - Guard the 1:1 set_* fallback: a multi-argument set_* with no handler now raises instead of silently dropping its extra arguments; add a regression test for that path. - Strengthen the export tests to assert the complete structures (matrix dtype/vehicle_type, both time-window arrays, capacity demand+capacity, and every field of both vehicle breaks including locations). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: Ramakrishna Prabhu <ramakrishnap@nvidia.com>
CI Test Summary✅ All 22 test job(s) passed. |
Summary
Adds
to_host_problem(dm): exports a recorded routing problem (the store-then-build IR from #1556) to host numpy arrays keyed by theRoutingProblemfields, copying any device (cuDF) inputs to host. This is the client-side foundation for serializing a routing problem over gRPC.Validated by a round-trip against the VRP
RoutingProblemproto; full routing suite passes.🤖 Generated with Claude Code