Fix hardcoded long int type

Before the type was incorrectly determined during the code generation
step, which prevented it from working cross-platform. Change that to
instead determine it when `numba_overloads.py` is imported.

numba_special/generate_overloads.py

@@ -24,6 +24,13 @@
 
 from . import function_pointers
 
+bytes_to_int_type = {{
+    2: numba.types.int16,
+    4: numba.types.int32,
+    8: numba.types.int64
+}}
+
+numba_long = bytes_to_int_type[ctypes.sizeof(ctypes.c_long)]
 
 functions = {{
 {FUNCTIONS}
@@ -71,7 +78,7 @@ def {FUNCTION}(*args):"""
 
 CTYPES_TO_NUMBA_TYPES = {
     'c_double': 'numba.types.float64',
-    'c_long': 'numba.types.int{}'.format(8 * ctypes.sizeof(ctypes.c_long))
+    'c_long': 'numba_long'
 }
 
 CTYPES_TO_SHORT_NUMBA_TYPES = {

numba_special/numba_overloads.py

@@ -6,6 +6,13 @@
 
 from . import function_pointers
 
+bytes_to_int_type = {
+    2: numba.types.int16,
+    4: numba.types.int32,
+    8: numba.types.int64
+}
+
+numba_long = bytes_to_int_type[ctypes.sizeof(ctypes.c_long)]
 
 functions = {
     'agm': ctypes.CFUNCTYPE(ctypes.c_double, ctypes.c_double, ctypes.c_double),
@@ -233,7 +240,7 @@ def bdtr(*args):
         f = get_scalar_function('bdtr[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('bdtr[long]')
         return lambda *args: f(*args)
 
@@ -244,7 +251,7 @@ def bdtrc(*args):
         f = get_scalar_function('bdtrc[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('bdtrc[long]')
         return lambda *args: f(*args)
 
@@ -255,7 +262,7 @@ def bdtri(*args):
         f = get_scalar_function('bdtri[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('bdtri[long]')
         return lambda *args: f(*args)
 
@@ -546,7 +553,7 @@ def eval_chebyc(*args):
         f = get_scalar_function('eval_chebyc[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_chebyc[long, double]')
         return lambda *args: f(*args)
 
@@ -557,7 +564,7 @@ def eval_chebys(*args):
         f = get_scalar_function('eval_chebys[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_chebys[long, double]')
         return lambda *args: f(*args)
 
@@ -568,7 +575,7 @@ def eval_chebyt(*args):
         f = get_scalar_function('eval_chebyt[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_chebyt[long, double]')
         return lambda *args: f(*args)
 
@@ -579,7 +586,7 @@ def eval_chebyu(*args):
         f = get_scalar_function('eval_chebyu[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_chebyu[long, double]')
         return lambda *args: f(*args)
 
@@ -590,7 +597,7 @@ def eval_gegenbauer(*args):
         f = get_scalar_function('eval_gegenbauer[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64, numba.types.float64,):
         f = get_scalar_function('eval_gegenbauer[long, double]')
         return lambda *args: f(*args)
 
@@ -601,21 +608,21 @@ def eval_genlaguerre(*args):
         f = get_scalar_function('eval_genlaguerre[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64, numba.types.float64,):
         f = get_scalar_function('eval_genlaguerre[long, double]')
         return lambda *args: f(*args)
 
 
 @numba.extending.overload(sc.eval_hermite)
 def eval_hermite(*args):
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_hermite')
         return lambda *args: f(*args)
 
 
 @numba.extending.overload(sc.eval_hermitenorm)
 def eval_hermitenorm(*args):
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_hermitenorm')
         return lambda *args: f(*args)
 
@@ -626,7 +633,7 @@ def eval_jacobi(*args):
         f = get_scalar_function('eval_jacobi[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64, numba.types.float64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64, numba.types.float64, numba.types.float64,):
         f = get_scalar_function('eval_jacobi[long, double]')
         return lambda *args: f(*args)
 
@@ -637,7 +644,7 @@ def eval_laguerre(*args):
         f = get_scalar_function('eval_laguerre[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_laguerre[long, double]')
         return lambda *args: f(*args)
 
@@ -648,7 +655,7 @@ def eval_legendre(*args):
         f = get_scalar_function('eval_legendre[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_legendre[long, double]')
         return lambda *args: f(*args)
 
@@ -659,7 +666,7 @@ def eval_sh_chebyt(*args):
         f = get_scalar_function('eval_sh_chebyt[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_sh_chebyt[long, double]')
         return lambda *args: f(*args)
 
@@ -670,7 +677,7 @@ def eval_sh_chebyu(*args):
         f = get_scalar_function('eval_sh_chebyu[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_sh_chebyu[long, double]')
         return lambda *args: f(*args)
 
@@ -681,7 +688,7 @@ def eval_sh_jacobi(*args):
         f = get_scalar_function('eval_sh_jacobi[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64, numba.types.float64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64, numba.types.float64, numba.types.float64,):
         f = get_scalar_function('eval_sh_jacobi[long, double]')
         return lambda *args: f(*args)
 
@@ -692,7 +699,7 @@ def eval_sh_legendre(*args):
         f = get_scalar_function('eval_sh_legendre[double, double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('eval_sh_legendre[long, double]')
         return lambda *args: f(*args)
 
@@ -745,7 +752,7 @@ def expn(*args):
         f = get_scalar_function('expn[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('expn[long]')
         return lambda *args: f(*args)
 
@@ -1078,7 +1085,7 @@ def kn(*args):
         f = get_scalar_function('kn[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('kn[long]')
         return lambda *args: f(*args)
 
@@ -1173,7 +1180,7 @@ def nbdtr(*args):
         f = get_scalar_function('nbdtr[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('nbdtr[long]')
         return lambda *args: f(*args)
 
@@ -1184,7 +1191,7 @@ def nbdtrc(*args):
         f = get_scalar_function('nbdtrc[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('nbdtrc[long]')
         return lambda *args: f(*args)
 
@@ -1195,7 +1202,7 @@ def nbdtri(*args):
         f = get_scalar_function('nbdtri[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba_long, numba.types.float64,):
         f = get_scalar_function('nbdtri[long]')
         return lambda *args: f(*args)
 
@@ -1325,7 +1332,7 @@ def pdtr(*args):
         f = get_scalar_function('pdtr[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('pdtr[long]')
         return lambda *args: f(*args)
 
@@ -1336,7 +1343,7 @@ def pdtrc(*args):
         f = get_scalar_function('pdtrc[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('pdtrc[long]')
         return lambda *args: f(*args)
 
@@ -1347,7 +1354,7 @@ def pdtri(*args):
         f = get_scalar_function('pdtri[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('pdtri[long]')
         return lambda *args: f(*args)
 
@@ -1428,7 +1435,7 @@ def smirnov(*args):
         f = get_scalar_function('smirnov[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('smirnov[long]')
         return lambda *args: f(*args)
 
@@ -1439,7 +1446,7 @@ def smirnovi(*args):
         f = get_scalar_function('smirnovi[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('smirnovi[long]')
         return lambda *args: f(*args)
 
@@ -1527,7 +1534,7 @@ def yn(*args):
         f = get_scalar_function('yn[double]')
         return lambda *args: f(*args)
 
-    if args == (numba.types.int64, numba.types.float64,):
+    if args == (numba_long, numba.types.float64,):
         f = get_scalar_function('yn[long]')
         return lambda *args: f(*args)