Primary goal
Re-establish one integrated constraint-solver line that reproduces the PhD results, covers other historical result families, and can stand in as the gold-standard codebase.
Public Project Page
Steven Woods
PhD Renovation
A living restoration of Steven Woods' PhD research code on automated program understanding via constraint satisfaction, rebuilt to run on modern SBCL while preserving the published thesis-result story and the broader historical solver lineage.
Current state
What is already solid
Version 1.0.0 ships with all four maintained historical solver snapshots loading on SBCL, the thesis ij2 / ij3 / ij4 validation path executable and CI-protected, the bounded AO baseline asserted, ff provenance checked, and both the older ADT archive and the earliest DREV-era T-series archive integrity-checkable.
What M1 means
M1 was the integration gate that made 1.0.0 possible: qcsp3 holds the PhD baseline, the asserted AO baseline, and the documented ff relationship under one supported executable story. The remaining archive-heavy families stay explicitly bounded rather than silently missing.
What comes next
The next major step after 1.0.0 is ingestion, de-duplication, and categorization of newly found code, data, publications, and experiment artifacts. ADT batch remains snapshot-specific, 1.0.1 is the right home for remaining STYLE-WARNING cleanup, and Hanoi-4 stays documented in 1.x while full behavior and repeatability understanding is deferred to 2.0.