APPENDIX — The Death of Classic OpenLane (v0.23)

This document exists for one reason:

To permanently close the discussion about “reviving Classic OpenLane”.

Classic OpenLane is not deprecated. It is not unsupported. It is irrecoverable.

This is not opinion. This is a postmortem.

What “Classic OpenLane” Refers To

In this document, Classic OpenLane means:

• OpenLane v0.23 era (2021–2022)
• Makefile-based flow
• Pre-volare PDK handling
• sky130A PDK distributed as tarballs
• efabless/openlane Docker images with embedded PDKs
• open_pdks build inside container or host

This environment was once stable. It is no longer reconstructible.

How Classic OpenLane Actually Died

Classic OpenLane did not die from bugs. It died from ecosystem erasure.

1. Docker Images Were Removed or Mutated

• efabless/openlane:v0.23 no longer exists as originally published
• Old digests were deleted
• Tags were repointed to volare-based images
• Images with embedded sky130A PDK are gone

Result:
You cannot pull a historically correct Classic container anymore.

2. sky130A PDK Tarballs Were Deleted

• GitHub Releases: removed
• S3 mirrors: removed
• Internal efabless mirrors: removed

Classic OpenLane depended on exact PDK snapshots. Those snapshots no longer exist.

Result:
Even if you had the flow, you cannot recreate the inputs.

3. open_pdks Became Unbuildable for Classic

• skywater-pdk now requires newer toolchains
• GLIBC requirements exceed Classic base images
• Conda/Miniconda assumptions changed
• Python tooling drifted

Classic OpenLane expected:

“Build PDK from open_pdks when needed”

That path is now broken.

4. Infrastructure Assumptions Changed

• ORFS legacy images removed
• “In-a-box” education containers removed
• Dependency pinning loosened upstream
• Continuous delivery replaced archival stability

Classic OpenLane assumed immutability. The ecosystem moved to continuous change.

Why “Just Rebuild It” Is Impossible

Attempts that do not work:

• Rebuilding open_pdks from current master
• Mixing old flow with new PDK
• Using volare with Classic flow
• Extracting PDKs from modern containers
• Rewriting paths or scripts
• Pinning random commits retroactively

These produce:

• Silent mismatches
• Subtle LVS/DRC differences
• STA drift
• GLS failures
• False confidence

A design that “runs” is not a design that is equivalent.

Responsibility Clarification

This Is NOT User Error

The loss of Classic OpenLane is due to:

• Upstream asset deletion
• Infrastructure reorganization
• Non-archival distribution practices

No amount of care by an end user could prevent this, unless they had offline full backups of:

• Docker images
• PDK tarballs
• Toolchains

That is not a reasonable expectation.

Why This Repository Exists Because of This

This repository is a direct response to this failure.

Lessons learned:

• Never depend on upstream permanence
• Never assume old versions can be rebuilt later
• Never rely on “I can always reinstall”
• Treat environments as artifacts, not setups

The Only Valid Replacement Strategy

Classic OpenLane cannot be restored.

What can be done:

• Use OpenLane1 (current Makefile-based) as a stable baseline
• Pin exact tool + PDK combinations
• Avoid rebuilding PDKs
• Avoid conda-based flows
• Use WSL export/import as the primary backup mechanism
• Freeze working environments

This reproduces functionality, not history.

That is sufficient. And that is the only realistic goal.

Final Verdict

Classic OpenLane is dead.

Not deprecated.
Not broken.
Not unsupported.

Gone.

Any guide suggesting otherwise is misleading you.

This repository exists so you never have to learn this the hard way again.