- [`KT-60574`](https://youtrack.jetbrains.com/issue/KT-60574) K2: generated IR for `suspendCoroutineUninterceptedOrReturn` is different from K1 (K2 uses Any? instead of Unit)
- [`KT-66512`](https://youtrack.jetbrains.com/issue/KT-66512) K2: Incorrect diagnostic in lambda whose expected type is a type alias to Unit

- github.com/google/cel-spec: [v0.6.0](https://github.com/google/cel-spec/tree/v0.6.0)
- github.com/jstemmer/go-junit-report: [v0.9.1](https://github.com/jstemmer/go-junit-report/tree/v0.9.1)
- github.com/nxadm/tail: [v1.4.4](https://github.com/nxadm/tail/tree/v1.4.4)
- github.com/onsi/ginkgo: [v1.14.0](https://github.com/onsi/ginkgo/tree/v1.14.0)

- E2e framework: added `--report-complete-ginkgo` and `--report-complete-junit` parameters. They work like `ginkgo --json-report <report dir>/ginkgo/report.json --junit-report <report dir>/ginkgo/report.xml`. ([#115678](https://github.com/kubernetes/kubernetes/pull/115678), [@pohly](https://github.com/pohly)) [SIG Testing]

  %18 = "tf.Identity"(%17) {device = ""} : (tensor<i1>) -> tensor<i1>
  %19 = "tf.StringLength"(%arg0) {device = "", unit = "BYTE"} : (tensor<1x!tf_type.string>) -> tensor<1xi32>
  %20 = "tf.ExpandDims"(%19, %7) {device = ""} : (tensor<1xi32>, tensor<i32>) -> tensor<1x1xi32>

  func.return %1 : tensor<4x4xf32>

  // CHECK-DAG: %[[unit:.*]] = "tfl.no_value"() <{value}> : () -> none
  // CHECK-DAG: %[[filter:.*]] = arith.constant dense<{{.*}}> : tensor<4x4xf32>
  // CHECK: %[[fc_result:.*]] = "tfl.fully_connected"(%arg0, %arg1, %[[unit]])
  // CHECK: %[[add_result:.*]] = tfl.add %[[fc_result]], %[[filter]]
  // CHECK: return %[[add_result]]
}

- If using cgroups v2, then the cgroup aware OOM killer will be enabled for container cgroups via  `memory.oom.group` .  This causes processes within the cgroup to be treated as a unit and killed simultaneously in the event of an OOM kill on any process in the cgroup. ([#117793](https://github.com/kubernetes/kubernetes/pull/117793), [@tzneal](https://github.com/tzneal)) [SIG Apps, Node and Testing]

    // reshape. Various code generation techniques benefit from the knowledge
    // that this is a restricted form of shape manipulation that is just adding
    // unit dims.
    if (op.getKeepDims()) {
      for (const auto &dim_is_reduced :
           llvm::enumerate(reduced_dimensions_bitmap)) {
        if (dim_is_reduced.value()) {
          auto index_attr = GetI32ElementsAttr(

d2=Rh(),Vh=Bh(),Uh=$h();function ft(i){return this instanceof ft?(this.nodes=d2(i),this):new ft(i)}ft.prototype.toString=function(){return Array.isArray(this.nodes)?Uh(this.nodes):""};ft.prototype.walk=function(i,e){return Vh(this.nodes,i,e),this};ft.unit=zh();ft.walk=Vh;ft.stringify=Uh;Wh.exports=ft});function Mo(i){return typeof i=="object"&&i!==null}function h2(i,e){let t=Ke(e);do if(t.pop(),(0,ri.default)(i,t)!==void 0)break;while(t.length);return t.length?t:void 0}function Vt(i){return typeof ...

	AllowInvalidPodDeletionCost bool
	// Allow invalid label-value in LabelSelector
	AllowInvalidLabelValueInSelector bool
	// Allow pod spec to use non-integer multiple of huge page unit size
	AllowIndivisibleHugePagesValues bool
	// Allow pod spec to use status.hostIPs in downward API if feature is enabled
	AllowHostIPsField bool

	shutdownOnce sync.Once
}

func (sc *http2serverConn) maxHeaderListSize() uint32 {
	n := sc.hs.MaxHeaderBytes
	if n <= 0 {
		n = DefaultMaxHeaderBytes
	}
	// http2's count is in a slightly different unit and includes 32 bytes per pair.
	// So, take the net/http.Server value and pad it up a bit, assuming 10 headers.
	const perFieldOverhead = 32 // per http2 spec
	const typicalHeaders = 10   // conservative