You can see the available versions (e.g. to check what is the current latest) in the [Release Notes](../release-notes.md){.internal-link target=_blank}.

## About versions

Following the Semantic Versioning conventions, any version below `1.0.0` could potentially add breaking changes.

FastAPI also follows the convention that any "PATCH" version change is for bug fixes and non-breaking changes.

!!! tip

* text eol=lf

#
# The above will handle all files NOT found below
# https://help.github.com/articles/dealing-with-line-endings/
# https://github.com/Danimoth/gitattributes

# These are explicitly windows files and should use crlf
*.bat           text eol=crlf

# These files are text and should be normalized (Convert crlf => lf)
*.bash          text eol=lf
*.css           text diff=css
*.htm           text diff=html

func DeclaredInMain() {
}

type C struct {
}

func F() *C {
	return nil
}

var slicePtr interface{} = &[]int{}

func main() {
	defer depBase.ImplementedInAsm()
	// This code below causes various go.itab.* symbols to be generated in
	// the executable. Similar code in ../depBase/dep.go results in
	// exercising https://golang.org/issues/17594
	reflect.TypeOf(os.Stdout).Elem()
	runtime.GC()

// CHECK:   description: "MLIR Converted.",
// CHECK:   buffers: [ {
// CHECK-EMPTY
// CHECK:   }, {
// CHECK-EMPTY
// CHECK:   } ]
// CHECK: }

func.func @main() -> tensor<*xi32> {
	// Tests that the below type attribute is convertible into the corresponding custom option in flatbuffer.
  %0 = "tf.SomeOperation"() {dtype = i32 } : () -> tensor<*xi32>
  func.return %0 : tensor<*xi32>

// Copyright 2020 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// The errors below are produced by the parser, but we check
// them here for consistency with the types2 tests.

package p

import ; // ERROR "missing import path"
import "" // ERROR "invalid import path (empty string)"
import
var /* ERROR "missing import path" */ _ int

// go into an infinite loop when handling negative dereference
// cycles. The critical thing being tested here is that compilation
// succeeds ("errorcheck -0"), not any particular diagnostic output,
// hence the very lax ERROR patterns below.

package p

type Node struct {
	Orig *Node
}

var sink *Node

func f1() {
	var n Node // ERROR "."
	n.Orig = &n

	m := n // ERROR "."
	sink = &m
}

func f2() {

			// copied to the slice. So if we append to the slice here none of the
			// updates we make below are reflected in the slice.
			newCondition = true
		}

		// Update the heartbeat time
		condition.LastHeartbeatTime = currentTime

		// Note: The conditions below take care of the case when a new NodeMemoryPressure condition is

			}
			v += int64(1 + m.rand(int(max)))
			return v
		case 1:
			// Subtract a random number
			if v <= -maxValue {
				continue
			}
			if v < 0 && maxValue+v < max {
				// Don't let v drop below -maxValue
				max = maxValue + v
			}
			v -= int64(1 + m.rand(int(max)))
			return v
		}
	}
}

func (m *mutator) mutateUInt(v, maxValue uint64) uint64 {
	var max uint64
	for {

limitations under the License.
==============================================================================*/
#include "tensorflow/cc/experimental/libtf/impl/string.h"

#include <unordered_set>

// It is important for the container below to not invalidate pointers to
// elements when elements are inserted, because the String class stores such
// pointers. This rules out, for example, absl::flat_hash_set.
using StringTable = std::unordered_set<std::string>;

# However, there are lots of commands the cache is not needed for, and we
# shouldn't require it when it won't be used.
#
# TODO(golang.org/issue/39882): commands below should work, too.
# * go clean -modcache
# * go env
# * go fix
# * go fmt
# * go generate
# * go get
# * go list (without -export or -compiled)

env GOCACHE=off