♻️ Separate target placement from routing

[burgholzer]

🤖 AI text below 🤖

Description

Separate deterministic qubit placement from topology-aware routing.

Target compilation now uses a compact placement pass for all-to-all targets and reserves the existing mapping pass for explicit coupling graphs. Both paths share the validated allocation discovery, capacity check, and dynamic-to-static rewrite, so the split does not duplicate the fragile part of the transformation.

This is the focused prerequisite for #2218. After this PR merges, #2218 can add its three-state connectivity policy without making the router handle an unknown topology: explicit connectivity uses mapping, all-to-all uses placement, and unknown connectivity rejects residual multi-site operations before placement.

Validated locally with:

• 86 mapping tests
• 139 compiler tests
• the generated MLIR documentation build
• the full repository lint suite
• the canonical changed-file C++ lint suite with no clang-format or clang-tidy findings

No changelog or upgrade-guide entry is needed because this only restructures the unreleased MQT Core v4 target-compilation work.

Checklist

• The pull request only contains commits that are focused and relevant to this change.
• I have added appropriate tests that cover the new/changed functionality.
• I have updated the documentation to reflect these changes.
• I have added entries to the changelog for any noteworthy additions, changes, fixes, or removals.
• I have added migration instructions to the upgrade guide (if needed).
• The changes follow the project's style guidelines and introduce no new warnings.
• The changes are fully tested and pass the CI checks.
• I have reviewed my own code changes.

If PR contains AI-assisted content:

• Any agent that created, edited, or submitted GitHub content was explicitly authorized for that scope, as required by our AI Usage Guidelines.
• Every agent-authored or agent-edited public text body begins with the visible disclosure 🤖 *AI text below* 🤖 (titles are exempt).
• I have disclosed AI assistance in the PR description.
• I confirm that I have personally reviewed and understood all AI-generated content, and accept full responsibility for it.

[codecov]

Codecov Report

❌ Patch coverage is 98.18182% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...lir/lib/Dialect/QCO/Transforms/Mapping/Mapping.cpp	98.1%	3 Missing ⚠️

📢 Thoughts on this report? Let us know!

[simon1hofmann]

Went through the changes and they look good to me 👍

I have one comment found by ponytail, where I'm not entirely sure if we should fix it that way:

[P2] The TableGen declaration creates an unsafe targetless public pass.

Passes.td generates createPlacementPass() and registerPlacementPass() with no CompilerTarget. That default-constructs PlacementPass, whose execution exits the process through reportFatalUsageError.

The canonical pipeline is safe because it calls the target-taking overload. The problem is the valid-looking public factory and generated place-qubits registration.

Ponytail fix: remove PlacementPass from TableGen and implement it as an explicitly target-bound PassWrapper, matching TargetNativeSynthesisPass.

Expose only:
createPlacementPass(const CompilerTarget&)

This removes code and an unusable API.

[coderabbitai]

Important

Approval pending

CodeRabbit has no unresolved comments, but it has not reviewed the latest commit.

Use the checkbox below to review the latest commit. CodeRabbit will approve the changes if it finds no blocking issues.

• 🔍 Trigger review

---

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.}

[MatthiasReumann]

Looks great! Pretty much the changes that I would expect.

Left some minor comments; otherwise let's get this merged!

[MatthiasReumann]

LGTM now 👍🏻