var reportFile: File?
        modifyState {
            val (newState, outputFile) = commitReportTo(outputDirectory, buildDisplayName, cacheAction, requestedTasks, totalProblemCount)
            reportFile = outputFile
            newState
        }
        return reportFile
    }

    private
    inline fun modifyState(f: State.() -> State) {

	// other values as these are modified.
	newState := baseState
	if updating {
		newState = blockLocs[b.ID].startState
	}

	for it := newState.Iterator(); !it.Done(); {
		k, d := it.Next()
		thisSlot := d.(*liveSlot)
		x := thisSlot.VarLoc
		x0 := x // initial value in newState

		// Intersect this slot with the slot in all the predecessors

      //   [H, W, ..., filter_in_depth, G, out_depth / G].
      auto new_shape = llvm::to_vector<6>(filter_shape);
      new_shape.back() = feature_group_count;
      new_shape.push_back(filter_shape.back() / feature_group_count);
      Type filter_element_ty = filter_ty.getElementType();
      auto ty =
          tensorflow::GetTypeFromTFTensorShape(new_shape, filter_element_ty);

      rewriter.replaceOp(op, {input});
      return success();
    }

    RankedTensorType shape_type =
        tensorflow::GetTypeFromTFTensorShape({-1}, rewriter.getIntegerType(32));
    auto new_shape = rewriter.create<TF::ShapeOp>(loc, shape_type, input);
    SmallVector<int64_t, 8> output_shape(/*Size=*/1, op.getNumElements());
    for (const auto &dim : dense_elem_attr.getValues<APInt>())

      SmallVector<int64_t> new_shape(shape.begin(), shape.end());
      // If dimension of the input type is dynamic. Update the
      // bounds of the dim with the new type if needed.
      for (int i = 0; i < input_ty.getShape().size(); i++) {
        if (hlo::isDynamicDimSize(input_ty.getShape()[i])) {
          new_bounds[i] = new_shape[i];
          new_shape[i] = ShapedType::kDynamic;
        }
      }

tensor.  The `input_indices` are recomputed based on the requested `new_shape`.

If one component of `new_shape` is the special value -1, the size of that
dimension is computed so that the total dense size remains constant.  At
most one component of `new_shape` can be -1.  The number of dense elements
implied by `new_shape` must be the same as the number of dense elements
originally implied by `input_shape`.