[RELAND] Remove some unnecessary <iostream> includes from headers (#108150)

In almost all cases this is only included for writing the output formatter, which
only uses `std::ostream` so including `<ostream>` is sufficient.

The istream header is ~1000 lines so the difference is non-trivial.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/108150
Approved by: https://github.com/albanD, https://github.com/malfet
ghstack dependencies: #108149

aten/src/ATen/DLConvertor.cpp

@@ -1,9 +1,6 @@
 #include <ATen/DLConvertor.h>
 #include <ATen/Functions.h>
 
-#include <iostream>
-#include <sstream>
-
 using namespace std;
 namespace at {
 

aten/src/ATen/autocast_mode.cpp

@@ -1,6 +1,5 @@
 #include <ATen/autocast_mode.h>
 
-#include <iostream>
 #include <exception>
 #include <mutex>
 #include <ATen/CachedTensorUtils.h>

aten/src/ATen/core/Range.cpp

@@ -1,6 +1,6 @@
 #include <ATen/core/Range.h>
 
-#include <iostream>
+#include <ostream>
 
 namespace at {
 

aten/src/ATen/core/dispatch/Dispatcher.h

@@ -18,6 +18,10 @@
 #include <ATen/core/grad_mode.h>
 #include <ATen/core/enum_tag.h>
 
+#ifndef NDEBUG
+#include <iostream>
+#endif
+
 namespace c10 {
 
 TORCH_API bool show_dispatch_trace();

aten/src/ATen/core/function_schema_inl.h

@@ -1,5 +1,6 @@
 #pragma once
-#include <iostream>
+#include <ostream>
+#include <sstream>
 
 // note: windows build doesn't find symbols in operator files unless
 // this is a header file

aten/src/ATen/core/interned_strings.cpp

@@ -4,7 +4,6 @@
 #include <ATen/core/interned_strings.h>
 #include <cstdint>
 #include <cstring>
-#include <iostream>
 #include <mutex>
 #include <sstream>
 #include <string>

aten/src/ATen/core/union_type.cpp

@@ -7,7 +7,8 @@
 #include <ATen/core/type_factory.h>
 #include <c10/macros/Macros.h>
 #include <c10/util/irange.h>
-#include <iostream>
+#include <ostream>
+#include <sstream>
 #include <utility>
 
 namespace c10 {

aten/src/ATen/cpu/vec/vec256/vec256_int.h

@@ -7,7 +7,6 @@
 #include <ATen/cpu/vec/vec_base.h>
 #include <c10/macros/Macros.h>
 #include <c10/util/irange.h>
-#include <iostream>
 
 namespace at {
 namespace vec {

aten/src/ATen/cpu/vec/vec256/vec256_qint.h

@@ -14,7 +14,6 @@
 
 #include <array>
 #include <cmath>
-#include <iostream>
 
 // This file defines Vectorized<> for the quantized types.
 //

aten/src/ATen/cpu/vec/vec512/vec512.h

@@ -18,7 +18,7 @@
 #include <cstddef>
 #include <cstdint>
 #include <cstring>
-#include <iostream>
+#include <ostream>
 
 namespace at {
 namespace vec {

aten/src/ATen/functorch/TensorWrapper.cpp

@@ -11,6 +11,8 @@
 #include <torch/library.h>
 #include <ATen/core/dispatch/Dispatcher.h>
 
+#include <iostream>
+
 namespace at {
 namespace functorch {
 

c10/util/logging_is_not_google_glog.h

@@ -4,9 +4,9 @@
 #include <chrono>
 #include <climits>
 #include <ctime>
-#include <fstream>
 #include <iomanip>
 #include <map>
+#include <ostream>
 #include <set>
 #include <sstream>
 #include <string>

functorch/csrc/dim/minpybind.h

@@ -8,7 +8,7 @@
 #define PY_SSIZE_T_CLEAN
 #include <Python.h>
 #include <utility>
-#include <iostream>
+#include <ostream>
 #include <memory>
 
 #define PY_BEGIN try {

torch/csrc/Module.cpp

@@ -28,6 +28,7 @@
 #include <torch/csrc/THConcat.h>
 #include <torch/csrc/utils/pybind.h>
 #include <cstdlib>
+#include <iostream>
 #include <unordered_map>
 
 #include <ATen/ThreadLocalPythonObjects.h>

torch/csrc/dynamo/compiled_autograd.h

@@ -4,7 +4,6 @@
 #include <torch/csrc/autograd/engine.h>
 #include <torch/csrc/utils/python_stub.h>
 #include <torch/csrc/utils/torch_dispatch_mode.h>
-#include <iostream>
 #include <typeindex>
 #include <vector>
 

torch/csrc/functorch/init.cpp

@@ -18,6 +18,8 @@
 #include <ATen/functorch/TensorWrapper.h>
 #include <c10/core/AutogradState.h>
 
+#include <iostream>
+
 // This file contains functorch's Python bindings.
 
 namespace torch {

torch/csrc/jit/api/module.cpp

@@ -19,6 +19,8 @@
 #include <torch/csrc/jit/passes/inliner.h>
 #include <torch/csrc/jit/runtime/operator.h>
 
+#include <iostream>
+
 namespace torch::jit {
 
 namespace {

torch/csrc/jit/codegen/fuser/codegen.h

@@ -6,9 +6,7 @@
 #include <torch/csrc/jit/codegen/fuser/tensor_desc.h>
 #include <torch/csrc/jit/ir/ir.h>
 
-#include <iostream>
 #include <string>
-#include <tuple>
 #include <vector>
 
 namespace torch {

torch/csrc/jit/codegen/fuser/tensor_desc.h

@@ -7,7 +7,7 @@
 #include <torch/csrc/Export.h>
 
 #include <algorithm>
-#include <iostream>
+#include <ostream>
 #include <vector>
 
 namespace torch {

torch/csrc/jit/frontend/concrete_module_type.cpp

@@ -3,6 +3,8 @@
 #include <c10/util/irange.h>
 #include <torch/csrc/jit/python/pybind_utils.h>
 
+#include <iostream>
+
 namespace torch::jit {
 
 ClassTypePtr ConcreteModuleTypeBuilder::createTypeFromThis() const {

torch/csrc/jit/frontend/source_range.h

@@ -3,10 +3,11 @@
 #include <c10/util/Optional.h>
 
 #include <algorithm>
-#include <iostream>
 #include <iterator>
 #include <memory>
 #include <numeric>
+#include <ostream>
+#include <sstream>
 #include <unordered_map>
 
 namespace torch {

torch/csrc/jit/frontend/tracer.h

@@ -12,7 +12,6 @@
 #include <torch/csrc/utils/variadic.h>
 
 #include <cstdint>
-#include <iostream>
 #include <memory>
 #include <mutex>
 #include <unordered_map>

torch/csrc/jit/ir/alias_analysis.cpp

@@ -10,6 +10,7 @@
 #include <torch/csrc/jit/runtime/operator.h>
 #include <torch/csrc/utils/memory.h>
 #include <fstream>
+#include <iostream>
 
 namespace torch::jit {
 

torch/csrc/jit/ir/ir.h

@@ -23,7 +23,7 @@
 #include <c10/util/Optional.h>
 
 #include <functional>
-#include <iostream>
+#include <iosfwd>
 #include <unordered_set>
 #include <vector>
 

torch/csrc/jit/jit_log.cpp

@@ -1,5 +1,6 @@
 #include <cstdlib>
 #include <iomanip>
+#include <iostream>
 #include <sstream>
 #include <string>
 #include <unordered_map>

torch/csrc/jit/mobile/model_tracer/MobileModelRunner.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include <iostream>
 #include <mutex>
 #include <sstream>
 

torch/csrc/jit/mobile/register_ops_common_utils.h

@@ -7,8 +7,6 @@
 #include <torch/csrc/jit/runtime/jit_exception.h>
 #include <torch/csrc/jit/runtime/vararg_functions.h>
 
-#include <iostream>
-
 namespace torch {
 namespace jit {
 

torch/csrc/jit/operator_upgraders/upgraders.h

@@ -1,7 +1,6 @@
 #pragma once
 #include <c10/macros/Export.h>
 #include <torch/csrc/jit/ir/ir.h>
-#include <iostream>
 #include <mutex>
 #include <string>
 #include <unordered_map>

torch/csrc/jit/operator_upgraders/upgraders_entry.h

@@ -1,11 +1,8 @@
 #pragma once
 #include <c10/macros/Export.h>
 #include <torch/csrc/jit/ir/ir.h>
-#include <iostream>
-#include <mutex>
 #include <string>
 #include <unordered_map>
-#include <vector>
 
 namespace torch {
 namespace jit {

torch/csrc/jit/operator_upgraders/utils.h

@@ -2,8 +2,6 @@
 #include <c10/macros/Export.h>
 #include <c10/util/Optional.h>
 #include <torch/csrc/jit/operator_upgraders/version_map.h>
-#include <iostream>
-#include <regex>
 #include <string>
 #include <vector>
 

torch/csrc/jit/passes/liveness.cpp

@@ -3,6 +3,7 @@
 #include <torch/csrc/jit/ir/alias_analysis.h>
 #include <torch/csrc/jit/ir/ir_views.h>
 #include <torch/csrc/jit/passes/constant_pooling.h>
+#include <iostream>
 #include <memory>
 
 namespace torch {

torch/csrc/jit/passes/onnx/pattern_conversion/pattern_conversion.cpp

@@ -7,6 +7,8 @@
 
 #include <ATen/ScalarOps.h>
 
+#include <iostream>
+
 // EDITING THIS FILE? READ THIS FIRST!
 // see Note [Edit Pattern Conversion] in pattern_conversion.h
 

torch/csrc/jit/passes/remove_inplace_ops.cpp

@@ -1,4 +1,5 @@
 #include <torch/csrc/jit/passes/remove_inplace_ops.h>
+#include <iostream>
 
 namespace torch {
 namespace jit {

torch/csrc/jit/runtime/argument_spec.cpp

@@ -1,6 +1,8 @@
 #include <c10/util/irange.h>
 #include <torch/csrc/jit/runtime/argument_spec.h>
 
+#include <iostream>
+
 namespace torch {
 namespace jit {
 

torch/csrc/jit/runtime/argument_spec.h

@@ -7,7 +7,7 @@
 #include <torch/csrc/Export.h>
 #include <torch/csrc/autograd/variable.h>
 #include <torch/csrc/jit/ir/ir.h>
-#include <iostream>
+#include <ostream>
 #include <vector>
 
 C10_CLANG_DIAGNOSTIC_PUSH()

torch/csrc/jit/runtime/print_handler.cpp

@@ -1,5 +1,6 @@
 #include <torch/csrc/jit/runtime/print_handler.h>
 
+#include <atomic>
 #include <iostream>
 #include <string>
 

torch/csrc/jit/runtime/print_handler.h

@@ -2,9 +2,7 @@
 
 #include <torch/csrc/Export.h>
 
-#include <atomic>
+#include <string>
-#include <functional>
-#include <iostream>
 
 namespace torch {
 namespace jit {

torch/csrc/jit/runtime/static/impl.cpp

@@ -28,6 +28,7 @@
 #include <torch/csrc/jit/runtime/vararg_functions.h>
 #include <algorithm>
 #include <cstdint>
+#include <iostream>
 
 #ifndef AT_PER_OPERATOR_HEADERS
 #include <ATen/NativeFunctions.h>

torch/csrc/jit/serialization/python_print.h

@@ -2,7 +2,6 @@
 #include <torch/csrc/Export.h>
 #include <torch/csrc/jit/api/module.h>
 #include <torch/csrc/jit/ir/ir.h>
-#include <iostream>
 #include <vector>
 
 namespace torch {

torch/csrc/jit/tensorexpr/bounds_inference.cpp

@@ -10,6 +10,8 @@
 
 #include <c10/util/irange.h>
 
+#include <iostream>
+
 namespace torch::jit::tensorexpr {
 
 using namespace analysis;

torch/csrc/jit/tensorexpr/bounds_overlap.cpp

@@ -3,6 +3,8 @@
 #include <torch/csrc/jit/tensorexpr/ir_visitor.h>
 #include <torch/csrc/jit/tensorexpr/stmt.h>
 
+#include <iostream>
+
 namespace torch::jit::tensorexpr::analysis {
 
 // Returns true if the given expression is guaranteed to be positive.

torch/csrc/jit/tensorexpr/ir_printer.cpp

@@ -6,6 +6,8 @@
 
 #include <c10/util/irange.h>
 
+#include <iostream>
+
 namespace torch::jit::tensorexpr {
 
 std::string IRPrinter::dtypeToCppString(const Dtype& dtype) {

torch/csrc/jit/tensorexpr/ir_printer.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include <iostream>
+#include <ostream>
 
 #include <torch/csrc/jit/tensorexpr/fwd_decls.h>
 #include <torch/csrc/jit/tensorexpr/ir.h>

torch/csrc/jit/tensorexpr/ir_verifier.h

@@ -1,7 +1,5 @@
 #pragma once
 
-#include <iostream>
-
 #include <torch/csrc/jit/tensorexpr/fwd_decls.h>
 #include <torch/csrc/jit/tensorexpr/ir_visitor.h>
 

torch/csrc/jit/tensorexpr/loopnest.cpp

@@ -1,6 +1,7 @@
 #include <torch/csrc/jit/tensorexpr/loopnest.h>
 
 #include <algorithm>
+#include <iostream>
 #include <stdexcept>
 #include <typeinfo>
 #include <unordered_map>

torch/csrc/jit/tensorexpr/loopnest_randomization.cpp

@@ -1,4 +1,5 @@
 #include <algorithm>
+#include <iostream>
 #include <random>
 #include <stdexcept>
 #include <typeinfo>

torch/csrc/jit/tensorexpr/mem_dependency_checker.cpp

@@ -3,6 +3,7 @@
 #include <c10/util/irange.h>
 
 #include <fstream>
+#include <iostream>
 
 namespace torch::jit::tensorexpr::analysis {
 

torch/csrc/jit/tensorexpr/registerizer.cpp

@@ -1,4 +1,5 @@
 #include <torch/csrc/jit/tensorexpr/registerizer.h>
+#include <iostream>
 
 namespace torch::jit::tensorexpr {
 namespace registerizer {

torch/csrc/jit/tensorexpr/types.h

@@ -1,7 +1,7 @@
 #pragma once
 
 #include <cstdint>
-#include <iostream>
+#include <iosfwd>
 
 #include <c10/core/ScalarType.h>
 #include <c10/util/Logging.h>

torch/csrc/lazy/core/debug_util.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include <iostream>
 #include <string>
 #include <vector>
 

torch/csrc/profiler/unwind/action.h

@@ -1,6 +1,6 @@
 #pragma once
 #include <stdint.h>
-#include <iostream>
+#include <ostream>
 
 enum {
   A_UNDEFINED = 0x0,

torch/csrc/profiler/unwind/eh_frame_hdr.h

@@ -1,6 +1,6 @@
 #pragma once
 #include <stdint.h>
-#include <iostream>
+#include <ostream>
 
 #include <torch/csrc/profiler/unwind/lexer.h>
 #include <torch/csrc/profiler/unwind/unwind_error.h>

torch/csrc/profiler/unwind/unwinder.h

@@ -2,7 +2,6 @@
 #include <stdint.h>
 #include <torch/csrc/profiler/unwind/action.h>
 #include <torch/csrc/profiler/unwind/unwind_error.h>
-#include <iostream>
 #include <limits>
 
 struct UnwindState {

torch/csrc/utils/throughput_benchmark.h

@@ -7,7 +7,7 @@
 
 #include <torch/csrc/jit/python/pybind_utils.h>
 
-#include <iostream>
+#include <iosfwd>
 #include <memory>
 #include <string>
 #include <vector>

torch/custom_class.h

@@ -12,7 +12,6 @@
 #include <c10/util/TypeTraits.h>
 #include <torch/custom_class_detail.h>
 #include <torch/library.h>
-#include <iostream>
 #include <sstream>
 
 namespace torch {

torch/lib/libshm/socket.h

@@ -9,8 +9,6 @@
 #include <cstddef>
 #include <cstdio>
 #include <cstring>
-#include <iostream>
-#include <sstream>
 #include <string>
 
 #include <libshm/alloc_info.h>