* Null row keys, columns keys, and values are not supported.
 *
 * <p>Lookups by row key are often faster than lookups by column key, because the data is stored in
 * a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs
 * quickly, since the row key is provided. However, {@code column(columnKey).size()} takes longer,
 * since an iteration across all row keys occurs.
 *

#if !defined(IS_MOBILE_PLATFORM)
  // Set up a multi-client distributed execution environment. Must be called on
  // all tasks in the cluster.
  // This call internally coordinates with other tasks to initialize the eager
  // context and TF server for multi-client execution.
  virtual absl::Status EnableCollectiveOps(const ServerDef& server_def) = 0;

   * When multiple elements are equivalent according to {@code compareTo()}, only the first one
   * specified is included. To create a copy of a {@code SortedSet} that preserves the comparator,
   * call {@link #copyOfSorted} instead. This method iterates over {@code elements} at most once.
   *
   * <p>Note that if {@code s} is a {@code Set<String>}, then {@code ImmutableSortedSet.copyOf(s)}

        <goals>
          <goal>process</goal>
        </goals>
      </execution>
    </executions>
  </plugin>
  .
  .
  .
</plugins>

You should do it after adding all your *path operations*.

```Python hl_lines="2  12-21  24"
{!../../docs_src/path_operation_advanced_configuration/tutorial002.py!}
```

/// tip

If you manually call `app.openapi()`, you should update the `operationId`s before that.

///

/// warning

If you do this, you have to make sure each one of your *path operation functions* has a unique name.

Just keep in mind that when you read "server" in general, it could refer to one of those two things.

When referring to the remote machine, it's common to call it **server**, but also **machine**, **VM** (virtual machine), **node**. Those all refer to some type of remote machine, normally running Linux, where you run programs.

## Install the Server Program

			staleWriters[i] = newBitrotWriter(disk, "", "testbucket", "testobject", erasure.ShardFileSize(test.size), test.algorithm, erasure.ShardSize())
		}

		// test case setup is complete - now call Heal()
		err = erasure.Heal(context.Background(), staleWriters, readers, test.size, nil)
		closeBitrotReaders(readers)
		closeBitrotWriters(staleWriters)
		if err != nil && !test.shouldFail {

    return result;
  }

  /*
   * For discussion of the safety of the following methods for operating on predecessors and
   * successors, see the comments near the end of CompactHashMap, noting that the methods here call
   * link(), which is defined at the end of this file.
   */

  private int getPredecessor(int entry) {
    return ((int) (link(entry) >>> 32)) - 1;
  }

  @Override
  int getSuccessor(int entry) {

	if offset%b.shardSize != 0 {
		// Offset should always be aligned to b.shardSize
		// Can never happen unless there are programmer bugs
		return 0, errUnexpected
	}
	if b.rc == nil {
		// For the first ReadAt() call we need to open the stream for reading.
		b.currOffset = offset
		streamOffset := (offset/b.shardSize)*int64(b.h.Size()) + offset
		if len(b.data) == 0 && b.tillOffset != streamOffset {

	// Refer to https://cloud.google.com/storage/docs/transcoding#decompressive_transcoding
	// Need to set `Accept-Encoding` header to `gzip` when issuing a GetObject call, to be able
	// to download the object in compressed state.
	// Calling ReadCompressed with true accomplishes that.
	object := gcs.client.Bucket(gcs.Bucket).Object(gcs.getDest(key)).ReadCompressed(true)