[XLA] Rename all (Mutable)ArraySlice to absl::Span.

PiperOrigin-RevId: 210998142

1 files changed, 9 insertions, 9 deletions

diff --git a/tensorflow/compiler/tests/randomized_tests.cc b/tensorflow/compiler/tests/randomized_tests.cc
index 1b8198dba8..0faf0fd8ed 100644
--- a/tensorflow/compiler/tests/randomized_tests.cc
+++ b/tensorflow/compiler/tests/randomized_tests.cc
@@ -275,13 +275,13 @@ class OpTest : public ::testing::Test {
 
   // Select a random element from 'candidates'.
   template <typename T>
-  T Choose(gtl::ArraySlice<T> candidates);
+  T Choose(absl::Span<const T> candidates);
 
   static constexpr int kDefaultMaxRank = 5;
   static constexpr int64 kDefaultMaxDimensionSize = 256LL;
 
   // Returns true if 'dims' have a size less than tf_xla_max_tensor_size.
-  bool TensorSizeIsOk(gtl::ArraySlice<int64> dims);
+  bool TensorSizeIsOk(absl::Span<const int64> dims);
 
   // Returns a random dimension size, in the range [min, max).
   int64 RandomDim(int64 min = 0, int64 max = kDefaultMaxDimensionSize);
@@ -307,11 +307,11 @@ class OpTest : public ::testing::Test {
   // of the type's range. If the shape is omitted, a random shape is used.
   // TODO(phawkins): generalize this code to a caller-supplied distribution.
   Tensor RandomTensor(DataType dtype, bool needs_unique_values,
-                      gtl::ArraySlice<int64> shape);
+                      absl::Span<const int64> shape);
   Tensor RandomTensor(DataType dtype);
 
   // Like RandomTensor, but uses values >= 0.
-  Tensor RandomNonNegativeTensor(DataType dtype, gtl::ArraySlice<int64> shape);
+  Tensor RandomNonNegativeTensor(DataType dtype, absl::Span<const int64> shape);
   Tensor RandomNonNegativeTensor(DataType dtype);
 
   // Returns a random subset of the integers in the range [0, rank), suitable
@@ -415,7 +415,7 @@ void OpTest::Repeatedly(const std::function<TestResult(void)>& fn) {
 }
 
 template <typename T>
-T OpTest::Choose(gtl::ArraySlice<T> candidates) {
+T OpTest::Choose(absl::Span<const T> candidates) {
   std::uniform_int_distribution<size_t> d(0, candidates.size() - 1);
   return candidates[d(generator())];
 }
@@ -425,7 +425,7 @@ int64 OpTest::RandomDim(int64 min, int64 max) {
   return size_distribution(generator());
 }
 
-bool OpTest::TensorSizeIsOk(gtl::ArraySlice<int64> dims) {
+bool OpTest::TensorSizeIsOk(absl::Span<const int64> dims) {
   int64 size = 1LL;
   for (int64 dim : dims) {
     size *= dim;
@@ -451,7 +451,7 @@ std::vector<int64> OpTest::RandomDims(int min_rank, int max_rank,
 }
 
 Tensor OpTest::RandomTensor(DataType dtype, bool needs_unique_values,
-                            gtl::ArraySlice<int64> shape) {
+                            absl::Span<const int64> shape) {
   Tensor tensor(dtype, TensorShape(shape));
   switch (dtype) {
     case DT_FLOAT: {
@@ -548,7 +548,7 @@ Tensor OpTest::RandomTensor(DataType dtype) {
 }
 
 Tensor OpTest::RandomNonNegativeTensor(DataType dtype,
-                                       gtl::ArraySlice<int64> shape) {
+                                       absl::Span<const int64> shape) {
   Tensor tensor(dtype, TensorShape(shape));
   switch (dtype) {
     case DT_FLOAT: {
@@ -1884,7 +1884,7 @@ TEST_F(OpTest, DynamicStitch) {
     for (int i = 0; i < n; ++i) {
       TensorShape shape(index_dims[i]);
       Tensor t = test::AsTensor<int32>(
-          gtl::ArraySlice<int32>(indices).subspan(pos, shape.num_elements()),
+          absl::Span<const int32>(indices).subspan(pos, shape.num_elements()),
           shape);
       builder.Input(t);
       pos += t.NumElements();