#include "tensorflow/core/platform/errors.h"

namespace tensorflow {
namespace gradients {
namespace {

// TODO(b/172558015): Using the pointer address as the identifier for the tensor
// may lead to collisions. Introduce another way to get a unique id for this
// tensor.
int64_t ToId(const AbstractTensorHandle* t) {
  return static_cast<int64_t>(reinterpret_cast<uintptr_t>(t));
}

   *       </ul>
   *   <li>another java.util.Set delegate implementation. In most modern JDKs, normal java.util hash
   *       collections intelligently fall back to a binary search tree if hash table collisions are
   *       detected. Rather than going to all the trouble of reimplementing this ourselves, we
   *       simply switch over to use the JDK implementation wholesale if probable hash flooding is

   * bulkGet(keys)} operation. See <a href="http://www.mathpages.com/home/kmath199.htm">Balls in
   * Bins model</a> for mathematical formulas that can be used to estimate the probability of
   * collisions.
   *
   * @param keys arbitrary non-null keys
   * @return the stripes corresponding to the objects (one per each object, derived by delegating to

   * Test that the hash function contains no funnels. A funnel is a situation where a set of input
   * (key) bits 'affects' a strictly smaller set of output bits. Funneling is bad because it can
   * result in more-than-ideal collisions for a non-uniformly distributed key space. In practice,
   * most key spaces are ANYTHING BUT uniformly distributed. A bit(i) in the input is said to

   *       </ul>
   *   <li>another java.util.Set delegate implementation. In most modern JDKs, normal java.util hash
   *       collections intelligently fall back to a binary search tree if hash table collisions are
   *       detected. Rather than going to all the trouble of reimplementing this ourselves, we
   *       simply switch over to use the JDK implementation wholesale if probable hash flooding is

   *       </ul>
   *   <li>another java.util.Map delegate implementation. In most modern JDKs, normal java.util hash
   *       collections intelligently fall back to a binary search tree if hash table collisions are
   *       detected. Rather than going to all the trouble of reimplementing this ourselves, we
   *       simply switch over to use the JDK implementation wholesale if probable hash flooding is

第二个版本是「绝对导入」：

```Python
from app.routers import items, users
```

要了解有关 Python 包和模块的更多信息，请查阅[关于 Modules 的 Python 官方文档](https://docs.python.org/3/tutorial/modules.html)。

///

### 避免名称冲突 { #avoid-name-collisions }

我们将直接导入 `items` 子模块，而不是仅导入其 `router` 变量。

这是因为我们在 `users` 子模块中也有另一个名为 `router` 的变量。

如果我们一个接一个地导入，例如：

```Python
from .routers.items import router
from .routers.users import router
```

   *       </ul>
   *   <li>another java.util.Map delegate implementation. In most modern JDKs, normal java.util hash
   *       collections intelligently fall back to a binary search tree if hash table collisions are
   *       detected. Rather than going to all the trouble of reimplementing this ourselves, we
   *       simply switch over to use the JDK implementation wholesale if probable hash flooding is

    }

    Ordering<Object> arbitrary =
        new ArbitraryOrdering() {
          @Override
          int identityHashCode(Object object) {
            return ((Integer) object) % 5; // fake tons of collisions!
          }
        };

    // Don't let the elements be in such a predictable order
    list = shuffledCopy(list, new Random(1));

    sort(list, arbitrary);

    }

    Ordering<Object> arbitrary =
        new ArbitraryOrdering() {
          @Override
          int identityHashCode(Object object) {
            return ((Integer) object) % 5; // fake tons of collisions!
          }
        };

    // Don't let the elements be in such a predictable order
    list = shuffledCopy(list, new Random(1));

    sort(list, arbitrary);