26600be023661a6606581b5e14cf939e90eea549,skll/data.py,,load_examples,#Any#Any#Any#Any#Any#,122

Before Change

                                             sparse))

    // Wait for processes to complete and store results
    pool.close()
    pool.join()
    ids = ids_result.get()
    classes = classes_result.get()
    features, feat_vectorizer = features_result.get()

After Change

    // memory (even though this requires reading the file multiple times).
    // Do this using a process pool so that we can clear out the temporary
    // variables more easily and do these in parallel.
    if MAX_CONCURRENT_PROCESSES == 1:
        ids = _ids_for_gen_func(example_gen_func, path, ids_to_floats)
        classes = _classes_for_gen_func(example_gen_func, path, tsv_label)
        features, feat_vectorizer = _features_for_gen_func(example_gen_func,
                                                           path, quiet, sparse)
    else:
        pool = Pool(min(3, MAX_CONCURRENT_PROCESSES))

        ids_result = pool.apply_async(_ids_for_gen_func,
                                      args=(example_gen_func, path,
                                            ids_to_floats))
        classes_result = pool.apply_async(_classes_for_gen_func,
                                          args=(example_gen_func, path,
                                                tsv_label))
        features_result = pool.apply_async(_features_for_gen_func,
                                           args=(example_gen_func, path, quiet,
                                                 sparse))

        // Wait for processes to complete and store results
        pool.close()
        pool.join()
        ids = ids_result.get()
        classes = classes_result.get()
        features, feat_vectorizer = features_result.get()

    return ExamplesTuple(ids, classes, features, feat_vectorizer)


def convert_examples(example_dicts, sparse=True, ids_to_floats=False):

Instances