You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avro.apache.org by "David Mollitor (Jira)" <ji...@apache.org> on 2020/04/15 14:47:00 UTC

[jira] [Created] (AVRO-2802) Pre-Size List in AvroInputFormat Avro File Lookup

David Mollitor created AVRO-2802:
------------------------------------

             Summary: Pre-Size List in AvroInputFormat Avro File Lookup
                 Key: AVRO-2802
                 URL: https://issues.apache.org/jira/browse/AVRO-2802
             Project: Apache Avro
          Issue Type: Improvement
            Reporter: David Mollitor
            Assignee: David Mollitor


{code:java}
    if (job.getBoolean(IGNORE_FILES_WITHOUT_EXTENSION_KEY, IGNORE_INPUTS_WITHOUT_EXTENSION_DEFAULT)) {
      List<FileStatus> result = new ArrayList<>();
      for (FileStatus file : super.listStatus(job))
        if (file.getPath().getName().endsWith(AvroOutputFormat.EXT))
          result.add(file);
      return result.toArray(new FileStatus[0]);
    } else {
      return super.listStatus(job);
    }
{code}

When a user runs an Avro MR job against a directory, it silently filters out files without an avro file extension. Fair enough.  However, anecdotally, this is the primary use scenario, so this code probably does not filter out many files.

I suggest that this {{ArrayList}} be pre-sized.  If there are a lot of files, and all of them have the avro file extension (base case), this {{ArrayList}} will had to be expanded multiple times (time and GC).  If there is a large list and it gets filtered down a lot, a few hundred bytes are wasted.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)