Fix HiGHs linking issue for python installs and minor rebase

.claude/LESSONS.md

@@ -110,6 +110,12 @@ Critical knowledge to avoid repeating mistakes.
 
 ## Python packaging / bindings
 
+- **A wheel repair tool does not protect local source/editable installs.**
+  HiGHS defaults to a shared library on Unix, leaving the installed Python
+  extension with `@rpath/libhighs.1.dylib` but no `LC_RPATH`; cibuildwheel can
+  repair release wheels, while `uv pip install -e .` cannot. Build HiGHS and
+  its extras as scoped PIC static libraries for Python packaging, and assert
+  the resulting CMake target type during configuration.
 - **Never infer Python import provenance from the checkout layout.** The
   2026-07-23 probe found a mixed environment: `dtwcpp` and `_api.py` resolve to
   `python/dtwcpp/` in the repository, while `_dtwcpp_core` resolves to the venv
@@ -327,6 +333,11 @@ Critical knowledge to avoid repeating mistakes.
 
 ## ARC SLURM Hardware
 
+- **macOS still ships Bash 3.2, where an empty array expansion fails under
+  `set -u`.** A Slurm script using `"${optional_args[@]}"` worked on newer
+  cluster Bash versions but exited with `optional_args[@]: unbound variable`
+  in local macOS tests. Build one always-nonempty command argument array and
+  conditionally append optional flags before expanding it.
 - **htc GPU compute capabilities (corrected from docs).** The ARC docs list CUDA toolkit version, not compute capability. Actual values: P100=6.0, V100=7.0, RTX8000/TitanRTX=7.5, A100=8.0, RTXA6000=8.6, L40S=8.9, H100/GH200=9.0.
 - **Rome (htc-g019) and Broadwell (htc-g045-049) lack AVX-512.** Use
   `DTWC_ARCH_LEVEL=v3` for a portable x86 htc build; there is no one x86-64-v4

CHANGELOG.md

@@ -8,6 +8,12 @@ This changelog contains a non-exhaustive list of new features and notable bug-fi
 <br/><br/>
 # Unreleased
 
+- Fixed macOS Python source and editable installs by statically bundling HiGHS
+  into the native extension, eliminating the unresolved
+  `@rpath/libhighs.1.dylib` import dependency. Made the generic Slurm job's
+  optional seed arguments compatible with macOS Bash 3.2, restored
+  platform-independent CLI discovery in Python tests, and included Matplotlib
+  in the Python test dependencies.
 - **Fix (build, GCC + LTO):** `run_openmp`'s exception-capture region is an
   unnamed `#pragma omp critical` again. The named form made GCC emit a COMMON
   `.gomp_critical_user_dtwc_run_openmp_exception` symbol into every TU that

cmake/Dependencies.cmake

@@ -39,22 +39,48 @@ function(dtwc_setup_dependencies)
   else()
     set(CUPDLP_GPU OFF CACHE BOOL "Enable HiGHS cuPDLP GPU support" FORCE)
   endif()
-  CPMAddPackage(
-    NAME highs
-    URL "https://github.com/ERGO-Code/HiGHS/archive/refs/tags/v1.15.1.tar.gz"
-    # SHA256 pinned (Task 0.12). Computed 2026-07-08 from the GitHub release
-    # tarball for the immutable tag v1.15.1 (`curl -sL … | sha256sum`).
-    URL_HASH SHA256=a840d269dff2fafb371dd247df13ad5e026d7ce3b35ad3dc1eedd59bf0c2fb16
-    SYSTEM
-    EXCLUDE_FROM_ALL
-    OPTIONS
-      "CI OFF"
-      "ZLIB OFF"
-      "BUILD_CXX_EXE OFF"
-      "BUILD_EXAMPLES OFF"
-      "BUILD_TESTING OFF"
-      "FAST_BUILD ON"
-    )
+  # A Python extension cannot rely on HiGHS' Unix default of BUILD_SHARED_LIBS=ON:
+  # the resulting @rpath/libhighs dependency is outside site-packages in local
+  # source/editable installs. Keep these normal variables scoped to the HiGHS
+  # subproject so native DTWC++ builds retain their requested shared-library
+  # policy. HiGHS sets PIC on its static target, making it safe to fold into the
+  # extension. GPU HiGHS is exempt because its Windows CUDA build requires DLLs.
+  block(SCOPE_FOR VARIABLES)
+    if(DTWC_BUILD_PYTHON AND NOT DTWC_HIGHS_GPU)
+      set(BUILD_SHARED_LIBS OFF)
+      set(BUILD_SHARED_EXTRAS_LIB OFF)
+    endif()
+
+    CPMAddPackage(
+      NAME highs
+      URL "https://github.com/ERGO-Code/HiGHS/archive/refs/tags/v1.15.1.tar.gz"
+      # SHA256 pinned (Task 0.12). Computed 2026-07-08 from the GitHub release
+      # tarball for the immutable tag v1.15.1 (`curl -sL … | sha256sum`).
+      URL_HASH SHA256=a840d269dff2fafb371dd247df13ad5e026d7ce3b35ad3dc1eedd59bf0c2fb16
+      SYSTEM
+      EXCLUDE_FROM_ALL
+      OPTIONS
+        "CI OFF"
+        "ZLIB OFF"
+        "BUILD_CXX_EXE OFF"
+        "BUILD_EXAMPLES OFF"
+        "BUILD_TESTING OFF"
+        "FAST_BUILD ON"
+      )
+  endblock()
+
+    # Fail during configuration, rather than shipping another extension with an
+    # unresolved libhighs dependency, if HiGHS changes how it honours
+    # BUILD_SHARED_LIBS.
+    if(DTWC_BUILD_PYTHON AND NOT DTWC_HIGHS_GPU AND TARGET highs)
+      get_target_property(_dtwc_highs_library_type highs TYPE)
+      if(NOT _dtwc_highs_library_type STREQUAL "STATIC_LIBRARY")
+        message(FATAL_ERROR
+          "Python packages require statically bundled HiGHS; got "
+          "${_dtwc_highs_library_type}")
+      endif()
+    endif()
+
     # Historically HiGHS <=1.14.0 had a debug assertion (ub_consistent) that
     # fired on valid warm-start MIP solves (primal-dual integral bookkeeping
     # tolerance 1e-12 too tight after a presolve reset). Retained defensively:

pyproject.toml

@@ -46,7 +46,7 @@ Issues = "https://github.com/Battery-Intelligence-Lab/dtw-cpp/issues"
 Changelog = "https://github.com/Battery-Intelligence-Lab/dtw-cpp/blob/main/CHANGELOG.md"
 
 [project.optional-dependencies]
-test = ["pytest>=6.0"]
+test = ["pytest>=6.0", "matplotlib>=3.7"]
 sklearn = ["scikit-learn>=1.0"]
 hdf5 = ["h5py>=3.0"]
 parquet = ["pyarrow>=12.0"]

scripts/slurm/jobs/cluster_generic.slurm

@@ -97,34 +97,35 @@ mkdir -p "${RESULTS_DIR}"
 
 T_START=$(date +%s.%N)
 DTWC_EXIT=0
-SEED_ARGS=()
+DTWC_ARGS=(
+    --input "${DTWC_INPUT}"
+    -k "${DTWC_K}"
+    --skip-cols "${DTWC_SKIP_COLS}"
+    --dtype "${DTWC_DTYPE}"
+    --method "${DTWC_METHOD}"
+    --band "${DTWC_BAND}"
+    --device "${DTWC_DEVICE}"
+    --max-iter "${DTWC_MAX_ITER}"
+    --variant "${DTWC_VARIANT}"
+    --wdtw-g "${DTWC_WDTW_G}"
+    --adtw-penalty "${DTWC_ADTW_PENALTY}"
+    --msm-c "${DTWC_MSM_C}"
+    --twe-nu "${DTWC_TWE_NU}"
+    --twe-lambda "${DTWC_TWE_LAMBDA}"
+    --mv-mode "${DTWC_MV_MODE}"
+    --missing-strategy "${DTWC_MISSING_STRATEGY}"
+    --metric "${DTWC_METRIC}"
+    --n-init "${DTWC_N_INIT}"
+)
 if [[ -n "${DTWC_SEED}" ]]; then
-    SEED_ARGS=(--seed "${DTWC_SEED}")
+    DTWC_ARGS+=(--seed "${DTWC_SEED}")
 fi
-"${DTWC_BIN}" \
-    --input "${DTWC_INPUT}" \
-    -k "${DTWC_K}" \
-    --skip-cols "${DTWC_SKIP_COLS}" \
-    --dtype "${DTWC_DTYPE}" \
-    --method "${DTWC_METHOD}" \
-    --band "${DTWC_BAND}" \
-    --device "${DTWC_DEVICE}" \
-    --max-iter "${DTWC_MAX_ITER}" \
-    --variant "${DTWC_VARIANT}" \
-    --wdtw-g "${DTWC_WDTW_G}" \
-    --adtw-penalty "${DTWC_ADTW_PENALTY}" \
-    --msm-c "${DTWC_MSM_C}" \
-    --twe-nu "${DTWC_TWE_NU}" \
-    --twe-lambda "${DTWC_TWE_LAMBDA}" \
-    --mv-mode "${DTWC_MV_MODE}" \
-    --missing-strategy "${DTWC_MISSING_STRATEGY}" \
-    --metric "${DTWC_METRIC}" \
-    --n-init "${DTWC_N_INIT}" \
-    "${SEED_ARGS[@]}" \
-    --name "${DTWC_NAME}" \
-    --output "${RESULTS_DIR}" \
-    --verbose \
-    || DTWC_EXIT=$?
+DTWC_ARGS+=(
+    --name "${DTWC_NAME}"
+    --output "${RESULTS_DIR}"
+    --verbose
+)
+"${DTWC_BIN}" "${DTWC_ARGS[@]}" || DTWC_EXIT=$?
 T_END=$(date +%s.%N)
 ELAPSED=$(echo "${T_END} - ${T_START}" | bc 2>/dev/null || echo "?")
 

tests/python/test_device.py

@@ -228,7 +228,24 @@ def test_no_python_side_copy_of_the_local_device(self):
     )
     @pytest.mark.parametrize(
         ("requested", "canonical"),
-        [("cuda", "gpu"), ("cuda:0", "gpu"), ("gpu:0", "gpu"), ("GPU", "gpu")],
+        [
+            pytest.param(
+                "cuda", "gpu",
+                marks=pytest.mark.skipif(
+                    not (dtwcpp.CUDA_AVAILABLE and dtwcpp.cuda_available()),
+                    reason="explicit CUDA unavailable",
+                ),
+            ),
+            pytest.param(
+                "cuda:0", "gpu",
+                marks=pytest.mark.skipif(
+                    not (dtwcpp.CUDA_AVAILABLE and dtwcpp.cuda_available()),
+                    reason="explicit CUDA unavailable",
+                ),
+            ),
+            ("gpu:0", "gpu"),
+            ("GPU", "gpu"),
+        ],
     )
     def test_gpu_aliases_canonicalise(self, requested, canonical):
         try:

tests/python/test_hpc.py

@@ -845,7 +845,8 @@ def test_restart_schedule_reaches_dtwc_cl(self):
         assert "DTWC_SEED=${SEED}" in wrapper
         assert 'DTWC_N_INIT="${DTWC_N_INIT:-1}"' in job
         assert '--n-init "${DTWC_N_INIT}"' in job
-        assert 'SEED_ARGS=(--seed "${DTWC_SEED}")' in job
+        assert 'DTWC_ARGS+=(--seed "${DTWC_SEED}")' in job
+        assert '"${DTWC_BIN}" "${DTWC_ARGS[@]}"' in job
         for name in (
             "MAX_ITER", "VARIANT", "WDTW_G", "ADTW_PENALTY", "MSM_C",
             "TWE_NU", "TWE_LAMBDA", "MV_MODE", "MISSING_STRATEGY", "METRIC",

tests/python/test_version_ssot.py

@@ -5,21 +5,20 @@
 import subprocess
 
 import dtwcpp
+from dtwcpp._hpc import find_dtwc_binary
 
 
 ROOT = Path(__file__).resolve().parents[2]
 
 
 def _cli_path() -> Path:
     override = os.environ.get("DTWC_CL_PATH")
-    candidates = [
-        Path(override) if override else None,
-        ROOT / "build" / "cfg-gate-normal" / "dtwc_cl.exe",
-        ROOT / "build" / "cfg-gate-normal" / "bin" / "dtwc_cl.exe",
-    ]
-    for candidate in candidates:
-        if candidate is not None and candidate.exists():
-            return candidate
+    if override and Path(override).exists():
+        return Path(override)
+
+    candidate = find_dtwc_binary(str(ROOT))
+    if candidate is not None:
+        return Path(candidate)
     raise AssertionError("dtwc_cl executable not found; set DTWC_CL_PATH")