Strip extra newlines from *.expected files

helion/_testing.py

@@ -209,6 +209,8 @@ def save(self) -> None:
                     f"--- assertExpectedJournal({name})\n{expected}\n\n"
                     for expected in expected_values
                 )
+            # Remove the last newline to play nicer with some people's editors
+            f.truncate(f.tell() - 1)
         os.rename(tmp, self.filename)
 
     @staticmethod

test/test_associative_scan.expected

@@ -1648,4 +1648,3 @@ def _test_cumsum_reverse_kernel_make_precompiler(x: torch.Tensor):
     _RDIM_SIZE_1 = triton.next_power_of_2(x.size(1))
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_test_cumsum_reverse_kernel_kernel)(x, result, x.size(1), result.stride(1), x.stride(1), _RDIM_SIZE_1, num_warps=4, num_stages=3)
-

test/test_atomic_add.expected

@@ -147,4 +147,3 @@ def _atomic_add_overlap_kernel_make_precompiler(x: torch.Tensor, y: torch.Tensor
     _BLOCK_SIZE_0 = 32
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_atomic_add_overlap_kernel_kernel)(indices, y, x, _BLOCK_SIZE_0, num_warps=4, num_stages=3)
-

test/test_broadcasting.expected

@@ -295,4 +295,3 @@ def _fn_make_precompiler(a, b):
     _BLOCK_SIZE_1 = 16
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_fn_kernel)(a, b, out, a.size(0), a.size(1), a.stride(0), a.stride(1), b.stride(0), out.stride(0), out.stride(1), _BLOCK_SIZE_0, _BLOCK_SIZE_1, num_warps=4, num_stages=3)
-

test/test_closures.expected

@@ -205,4 +205,3 @@ def _call_func_arg_on_host_make_precompiler(a, alloc):
     _BLOCK_SIZE_0 = 512
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_call_func_arg_on_host_kernel)(a, out, a.size(0), a.stride(0), out.stride(0), _BLOCK_SIZE_0, num_warps=4, num_stages=3)
-

test/test_constexpr.expected

@@ -116,4 +116,3 @@ def _fn_make_precompiler(x: torch.Tensor, s: hl.constexpr):
     _BLOCK_SIZE_1 = 16
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_fn_kernel)(x, out, out.stride(0), out.stride(1), x.stride(0), b, _BLOCK_SIZE_0, _BLOCK_SIZE_1, num_warps=4, num_stages=3)
-

test/test_control_flow.expected

@@ -214,4 +214,3 @@ def _fn_make_precompiler(x: torch.Tensor, y: torch.Tensor):
     output = torch.zeros_like(x)
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_fn_kernel)(x, y, output, output.stride(0), x.stride(0), y.stride(0), num_warps=4, num_stages=3)
-

test/test_generate_ast.expected

@@ -548,4 +548,3 @@ def _torch_ops_pointwise_make_precompiler(x, y):
     _BLOCK_SIZE_0 = 128
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_torch_ops_pointwise_kernel)(x, y, out, x.size(0), out.stride(0), x.stride(0), y.stride(0), _BLOCK_SIZE_0, num_warps=4, num_stages=3)
-

test/test_indexing.expected

@@ -171,4 +171,3 @@ def _pairwise_add_make_precompiler(x: torch.Tensor):
     _BLOCK_SIZE_0 = 32
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_pairwise_add_kernel)(out, x, out.size(0), out.stride(0), x.stride(0), _BLOCK_SIZE_0, num_warps=4, num_stages=3)
-

test/test_reduce.expected

@@ -590,4 +590,3 @@ def _test_reduce_keep_dims_kernel_make_precompiler(x: torch.Tensor):
     _RDIM_SIZE_1 = triton.next_power_of_2(x.size(1))
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_test_reduce_keep_dims_kernel_kernel)(x, result, x.size(0), x.size(1), result.stride(0), x.stride(0), x.stride(1), _BLOCK_SIZE_0, _RDIM_SIZE_1, num_warps=4, num_stages=3)
-

test/test_specialize.expected

@@ -184,4 +184,3 @@ def _fn_make_precompiler(x: torch.Tensor):
     _BLOCK_SIZE_0_1 = 32
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_fn_kernel)(x, out, x.size(0), x.size(1), out.stride(0), out.stride(1), x.stride(0), x.stride(1), scale, _BLOCK_SIZE_0_1, num_warps=4, num_stages=3)
-

test/test_type_propagation.expected

@@ -822,4 +822,3 @@ def root_graph_0():
     out: "i32[s77]" = helion_language__tracing_ops__host_tensor('out')
     store = helion_language_memory_ops_store(out, [block_size_0], convert_element_type, None);  out = block_size_0 = convert_element_type = store = None
     return None
-

test/test_views.expected

@@ -117,4 +117,3 @@ def _fn_make_precompiler(x: torch.Tensor, y: torch.Tensor):
     _BLOCK_SIZE_1 = 32
     from helion.runtime.precompile_shim import make_precompiler
     return make_precompiler(_fn_kernel)(x, y, out, out.size(0), out.size(1), x.size(0), x.size(1), y.size(0), out.stride(0), out.stride(1), x.stride(0), x.stride(1), y.stride(0), y.stride(1), _BLOCK_SIZE_0, _BLOCK_SIZE_1, num_warps=4, num_stages=3)
-