fix compile error

Author: joocer

opteryx/__version__.py

@@ -1,9 +1,9 @@
 # THIS FILE IS AUTOMATICALLY UPDATED DURING THE BUILD PROCESS
 # DO NOT EDIT THIS FILE DIRECTLY
 
-__build__ = 1857
+__build__ = 1861
 __author__ = "@joocer"
-__version__ = "0.26.2-beta.1857"
+__version__ = "0.26.2-beta.1861"
 
 # Store the version here so:
 # 1) we don't load dependencies by storing it in __init__.py

opteryx/draken/vectors/vector.pxd

@@ -15,7 +15,7 @@ cdef const uint64_t MIX_HASH_CONSTANT
 cdef const uint64_t NULL_HASH
 
 cdef extern from "simd_hash.h":
-    void simd_mix_hash(uint64_t* dest, const uint64_t* values, size_t count, uint64_t mix_constant) nogil
+    void simd_mix_hash(uint64_t* dest, const uint64_t* values, size_t count) nogil
 
 cdef inline uint64_t mix_hash(uint64_t current, uint64_t value) nogil:
     cdef uint64_t mixed = current ^ value

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "opteryx"
-version = "0.26.2-beta.1857"
+version = "0.26.2-beta.1861"
 description = "Query your data, where it lives"
 requires-python = '>=3.11'
 readme = {file = "README.md", content-type = "text/markdown"}

src/cpp/simd_hash.cpp

@@ -11,10 +11,10 @@
 
 namespace {
 
-inline void scalar_mix(uint64_t* dest, const uint64_t* values, std::size_t count, uint64_t mix_constant) {
+inline void scalar_mix(uint64_t* dest, const uint64_t* values, std::size_t count) {
     for (std::size_t i = 0; i < count; ++i) {
         uint64_t mixed = dest[i] ^ values[i];
-        mixed *= mix_constant;
+        mixed *= MIX_HASH_CONSTANT;
         mixed ^= mixed >> 32;
         dest[i] = mixed;
     }
@@ -53,14 +53,14 @@ inline uint64x2_t mullo_u64(uint64x2_t a, uint64x2_t b) {
 
 }  // namespace
 
-void simd_mix_hash(uint64_t* dest, const uint64_t* values, std::size_t count, uint64_t mix_constant) {
+void simd_mix_hash(uint64_t* dest, const uint64_t* values, std::size_t count) {
     if (dest == nullptr || values == nullptr || count == 0) {
         return;
     }
 
 #if defined(__AVX2__)
     const std::size_t stride = 4;
-    const __m256i const_vec = _mm256_set1_epi64x(static_cast<long long>(mix_constant));
+    const __m256i const_vec = _mm256_set1_epi64x(static_cast<long long>(MIX_HASH_CONSTANT));
     std::size_t i = 0;
     for (; i + stride <= count; i += stride) {
         __m256i dst_vec = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(dest + i));
@@ -72,11 +72,11 @@ void simd_mix_hash(uint64_t* dest, const uint64_t* values, std::size_t count, ui
         _mm256_storeu_si256(reinterpret_cast<__m256i*>(dest + i), combined);
     }
     if (i < count) {
-        scalar_mix(dest + i, values + i, count - i, mix_constant);
+        scalar_mix(dest + i, values + i, count - i);
     }
 #elif defined(__ARM_NEON) || defined(__ARM_NEON__)
     const std::size_t stride = 2;
-    const uint64x2_t const_vec = vdupq_n_u64(mix_constant);
+    const uint64x2_t const_vec = vdupq_n_u64(MIX_HASH_CONSTANT);
     std::size_t i = 0;
     for (; i + stride <= count; i += stride) {
         uint64x2_t dst_vec = vld1q_u64(dest + i);
@@ -88,9 +88,9 @@ void simd_mix_hash(uint64_t* dest, const uint64_t* values, std::size_t count, ui
         vst1q_u64(dest + i, combined);
     }
     if (i < count) {
-        scalar_mix(dest + i, values + i, count - i, mix_constant);
+        scalar_mix(dest + i, values + i, count - i);
     }
 #else
-    scalar_mix(dest, values, count, mix_constant);
+    scalar_mix(dest, values, count);
 #endif
 }

src/cpp/simd_hash.h

@@ -7,7 +7,12 @@
 extern "C" {
 #endif
 
-void simd_mix_hash(uint64_t* dest, const uint64_t* values, size_t count, uint64_t mix_constant);
+// Shared mixing constant used by scalar and SIMD mixers.
+#ifndef MIX_HASH_CONSTANT
+#define MIX_HASH_CONSTANT ((uint64_t)0x9e3779b97f4a7c15ULL)
+#endif
+
+void simd_mix_hash(uint64_t* dest, const uint64_t* values, size_t count);
 
 #ifdef __cplusplus
 }

tests/draken/morsels/test_hash_into.py

@@ -34,6 +34,12 @@ def _mix_hash(current: int, value: int, mix_constant: int) -> int:
 
 
 def _vector_hash_to_list(vector, mix_constant: int = 0x9E3779B97F4A7C15):
+    """Compute vector hashes into a Python list.
+
+    The `mix_constant` parameter is accepted for compatibility with older
+    tests but is ignored because the mixing constant is now enforced
+    internally by the implementation.
+    """
     length = getattr(vector, "length", None)
 
     if length is None:
@@ -43,7 +49,7 @@ def _vector_hash_to_list(vector, mix_constant: int = 0x9E3779B97F4A7C15):
             length = len(vector.to_pylist())
 
     out = array("Q", [0] * length)
-    hash_into_vector(vector, out, mix_constant=mix_constant)
+    hash_into_vector(vector, out)
     return list(out)
 
 
@@ -317,7 +323,7 @@ def test_hash_into_consistency_with_hash():
         expected.append(hashed)
 
     out_buf = array('Q', [0] * len(expected))
-    hash_into_vector(vec, out_buf, 0, mix_constant)
+    hash_into_vector(vec, out_buf, 0)
 
     assert list(out_buf) == expected
 
@@ -750,8 +756,8 @@ def test_hash_into_zero_mix_constant():
     out_buf = array('Q', [0, 0, 0])
     out_view = memoryview(out_buf)
 
-    # Should not raise even with zero mix constant
-    hash_into_vector(vec, out_view, 0, 0)
+    # Should not raise (mix constant override removed)
+    hash_into_vector(vec, out_view, 0)
 
     result = list(out_buf)
     baseline = _vector_hash_to_list(vec, mix_constant=MIX_HASH_CONSTANT)
@@ -769,7 +775,7 @@ def test_hash_into_max_uint64_mix_constant():
     out_view = memoryview(out_buf)
 
     max_uint64 = 0xFFFFFFFFFFFFFFFF
-    hash_into_vector(vec, out_view, 0, max_uint64)
+    hash_into_vector(vec, out_view, 0)
 
     result = list(out_buf)
     assert all(h != 0 for h in result)

third_party/mabel/draken/vectors/array_vector.pyx

@@ -373,7 +373,7 @@ cdef class ArrayVector(Vector):
                 encoded = repr(value).encode("utf-8")
                 scratch[j] = hash_bytes(encoded)
 
-            simd_mix_hash(&out_buf[offset + i], scratch_ptr, <size_t> block, MIX_HASH_CONSTANT)
+            simd_mix_hash(&out_buf[offset + i], scratch_ptr, <size_t> block)
             i += block
 
     def __str__(self):

third_party/mabel/draken/vectors/bool_vector.pyx

@@ -303,7 +303,7 @@ cdef class BoolVector(Vector):
                         scratch[j] = TRUE_HASH
                     else:
                         scratch[j] = FALSE_HASH
-                simd_mix_hash(dst + i, scratch_ptr, <size_t> block, MIX_HASH_CONSTANT)
+                simd_mix_hash(dst + i, scratch_ptr, <size_t> block)
                 i += block
             return
 

third_party/mabel/draken/vectors/date32_vector.pyx

@@ -347,7 +347,7 @@ cdef class Date32Vector(Vector):
                     block = DATE32_HASH_CHUNK
                 for j in range(block):
                     scratch[j] = <uint64_t>(<int64_t> data[i + j])
-                simd_mix_hash(dst + i, scratch_ptr, <size_t> block, MIX_HASH_CONSTANT)
+                simd_mix_hash(dst + i, scratch_ptr, <size_t> block)
                 i += block
             return
 

third_party/mabel/draken/vectors/float64_vector.pyx

@@ -346,7 +346,6 @@ cdef class Float64Vector(Vector):
         self,
         uint64_t[::1] out_buf,
         Py_ssize_t offset=0,
-        uint64_t mix_constant=<uint64_t>0x9e3779b97f4a7c15U,
     ) except *:
         cdef DrakenFixedBuffer* ptr = self.ptr
         cdef double* data = <double*> ptr.data
@@ -365,10 +364,7 @@ cdef class Float64Vector(Vector):
         cdef uint8_t byte
         cdef uint64_t value
 
-        mix_constant = MIX_HASH_CONSTANT  # enforce shared mixing constant
-        if mix_constant != MIX_HASH_CONSTANT:
-            mix_constant = MIX_HASH_CONSTANT
-
+        # Use shared MIX_HASH_CONSTANT directly; no need to pass it in.
         if has_nulls:
             for i in range(n):
                 byte = null_bitmap[i >> 3]
@@ -378,7 +374,7 @@ cdef class Float64Vector(Vector):
                     value = NULL_HASH
                 dst[i] = mix_hash(dst[i], value)
         else:
-            simd_mix_hash(dst, bits, <size_t>n, mix_constant)
+            simd_mix_hash(dst, bits, <size_t>n)
             return
 
     def __str__(self):