Extract compute_slice_from_label from compute_failure_slices

[mikeagun]

Refactor backward slicing to expose a general-purpose compute_slice_from_label() method on AnalysisResult. It accepts an arbitrary label and seed relevance set, making backward slicing available for any program point — not just failure labels.

The existing compute_failure_slices() now delegates to it, iterating error labels and extracting initial relevance from assertions. The slicing algorithm is unchanged.

Files changed: src/result.hpp, src/result.cpp

Summary by CodeRabbit

Release Notes

• Refactor
  • Restructured failure-slice computation to improve code organization and decouple slice generation from global iteration logic.
  • Slice construction now properly initializes error information on a per-label basis.
  • Introduced new compute_slice_from_label() API method for computing individual failure slices with customizable parameters.

[coderabbitai]

Walkthrough

Refactored compute_failure_slices to extract its backward-slicing logic into a new helper function compute_slice_from_label, which now accepts a target label and seed relevance. The public method delegates to this helper per error label. Updated slice initialization to derive .error from invariants[label].error when present, otherwise use empty VerificationError("").

Changes

Cohort / File(s)	Summary
Public API Addition src/result.hpp	Declared new public method compute_slice_from_label with parameters for program, label, seed relevance, and optional max_steps (default 200).
Backward-Slicing Refactoring src/result.cpp	Extracted inlined backward-slicing logic from compute_failure_slices into new compute_slice_from_label helper. Refactored slice construction to initialize .error from invariants[label] instead of copying from loop. Adjusted variable renaming (parents_of_target) and loop-guard logic while preserving worklist deduplication and join-point expansion mechanics.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~30 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main refactoring: extracting a new compute_slice_from_label method from the existing compute_failure_slices implementation.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/result.cpp`:
- Around line 632-640: The fallback currently repopulates
initial_relevance.registers from all assertions when either no failing assertion
was found or when the failing assertion has no register deps, which prevents
compute_slice_from_label() from receiving an empty seed (used to enter
conservative mode); change the logic so that aggregation from
extract_assertion_registers(assertion) only happens when !found_failing, and if
found_failing is true but initial_relevance.registers is empty, leave the seed
empty (do not insert any registers), preserving the conservative empty-seed
behavior.
- Around line 537-545: The per-label relevance is being overwritten when
revisiting a label; instead of assigning slice_labels[current_label] =
relevant_before (in both the instruction_contributes branch and the "No deps
available" branch where relevant_before = relevant_after), merge the new
relevance with any existing relevance for that label. Change these writes to
union/merge operations that combine existing slice_labels[current_label] with
relevant_before (using the same union semantics as visited), so additional
registers/stack slots from later visits are not lost; keep using the same
symbols: instruction_contributes, relevant_before, relevant_after, slice_labels,
and visited.

🪄 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: Repository UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: d98cc6e2-4bfe-43ac-97ad-e7003de4a933

📥 Commits

Reviewing files that changed from the base of the PR and between ce026ec and eb4e75a.

📒 Files selected for processing (2)

• src/result.cpp
• src/result.hpp