}

			// check that we got all violations
			t.Logf("Got violations: %q", cond.Message)
			for _, v := range tst.expectedViolations {
				if strings.Index(cond.Message, v) == -1 {
					t.Errorf("expected violation %q, but didn't get it", v)
				}
			}
			for _, v := range tst.unexpectedViolations {
				if strings.Index(cond.Message, v) != -1 {
					t.Errorf("unexpected violation %q", v)
				}
			}
		})

        "@com_google_absl//absl/types:optional",
        "@local_xla//xla:parse_flags_from_env",
    ],
)

# Header-only version of "flags" library, for linking from the shared object
# without ODR violations.
cc_library(
    name = "flags_headers",
    hdrs = ["flags.h"],
    visibility = [":friends"],
    deps = [
        "//tensorflow/core:framework_internal",
        "//tensorflow/core:framework_types_hdr",

# TODO: remove those flags when they are flipped in the default Bazel version TF uses.
build --incompatible_enforce_config_setting_visibility
# TODO: also enable this flag after fixing the visibility violations
# build --incompatible_config_setting_private_default_visibility

# Default options should come above this line.

# Android configs. Bazel needs to have --cpu and --fat_apk_cpu both set to the

Each of these requirements is treated as a configuration cache "problem" and fails the build if violations are present.

For the most part these requirements are actually surfacing some undeclared inputs.
In other words, using the configuration cache is an opt-in to more strictness, correctness and reliability for all builds.

   // OK
   %0 = "tf.Conv2D"(%arg0, %arg1) {T = "tfdtype$DT_FLOAT", data_format = "NHWC", dilations = [1, 2, 3, 1], padding = "SAME", strides = [1, 4, 5, 1]} : (tensor<256x32x32x3xf32>, tensor<3x3x3x16xf32>) -> tensor<256x8x7x16xf32>
   // Unsupported data format

        mean, variance = scale, offset
        out = nn_ops.conv2d(
            input_tensor,
            self.filters,
            strides=strides,
            dilations=dilations,
            padding=padding,
            data_format='NHWC',
        )
        if has_bias:
          out = nn_ops.bias_add(out, self.bias, data_format='NHWC')
        if has_batch_norm:

    has_batch_norm = True
    dilations = None
    input_shape = (1, 3, 4, 3)
    filter_shape = (2, 3, 3, 2)
    strides = (1, 1, 1, 1)
    model = self._create_conv2d_model(
        input_shape,
        filter_shape,
        self._input_saved_model_path,
        bias_fn,
        activation_fn,
        has_batch_norm,
        strides,
        dilations,
    )

    %0 = "tf.Conv2D"(%arg0, %arg1) {attr_map = "0:strides,1:use_cudnn_on_gpu,2:padding,3:explicit_paddings,4:dilations", data_format = "NHWC", device = "", dilations = [1, 1, 1, 1], explicit_paddings = [], padding = "VALID", strides = [1, 1, 2, 1], use_cudnn_on_gpu = true} : (tensor<*xf32>, tensor<*xf32>) -> tensor<*xf32>

    %2 = "tf.Cast"(%identity) {Truncate = false} : (tensor<2x3x3x2xi8>) -> tensor<2x3x3x2xi32>
    %3 = "tf.Sub"(%2, %arg6) : (tensor<2x3x3x2xi32>, tensor<i32>) -> tensor<2x3x3x2xi32>
    %4 = "tf.Conv2D"(%1, %3) {dilations = [1, 1, 1, 1], explicit_paddings = [], padding = "SAME", strides = [1, 1, 2, 1], use_cudnn_on_gpu = true} : (tensor<1x3x4x3xi32>, tensor<2x3x3x2xi32>) -> tensor<1x3x2x2xi32>

    ConvertTFConvOpMatchState state;
    state.stride_height = height;
    state.stride_width = width;

    if (TFIntListIs1XY1(op, "dilations", &height, &width)) {
      state.dilation_height_factor = height;
      state.dilation_width_factor = width;
    } else {
      // If the 'dilations' attribute is missing, we use the default value (1)
      // for both dilation height and width factor.