xai-org/x-algorithm · error · IllegalArgumentException

--$preferred (or deprecated --$deprecated) must be an intege

Error message

--$preferred (or deprecated --$deprecated) must be an integer; got '$s'

What it means

UnderTheHoodCommon.preferredInt parses a command-line arg (trying the preferred flag, then a deprecated alias) with s.toInt and throws when parsing fails. It exists to migrate users off a deprecated flag name while accepting either.

Source

Thrown at under-the-hood/scalding/UnderTheHoodCommon.scala:128

  def parseUserIds(args: Args): Set[Long] = {
    val raw = args.list("userIds").flatMap(_.split(",")).map(_.trim).filter(_.nonEmpty)
    val invalid = raw.filter(s => Try(s.toLong).isFailure)
    require(
      invalid.isEmpty,
      s"--userIds supplied but contains non-numeric ids: ${invalid.mkString(", ")}. " +
        "Omit --userIds entirely for an all-user run."
    )
    raw.flatMap(s => Try(s.toLong).toOption).toSet
  }

  def preferredInt(args: Args, preferred: String, deprecated: String, default: Int): Int = {
    val raw = args.optional(preferred).orElse(args.optional(deprecated))
    raw match {
      case None => default
      case Some(s) =>
        Try(s.toInt).getOrElse {
          throw new IllegalArgumentException(
            s"--$preferred (or deprecated --$deprecated) must be an integer; got '$s'")
        }
    }
  }

  def inScope(testUserIds: Set[Long], userId: Long): Boolean =
    testUserIds.isEmpty || testUserIds.contains(userId)

  def applyReducers[K, V](g: Grouped[K, V], reducers: Int): Grouped[K, V] =
    if (reducers > 0) g.withReducers(reducers) else g
}

View on GitHub (pinned to 24c60942c5)

Solutions

1. Re-run with a plain integer for the preferred (or deprecated) flag, e.g. --preferred 100
2. Check shell scripts/cron templates for stray characters or unreplaced placeholders in that arg
3. Drop the flag entirely to accept the documented default
4. If you intended unit suffixes (k/m), expand them to plain integers

Example fix

# before
hadoop jar job.jar com.twitter.uth.Job --topUsers 10x
# after
hadoop jar job.jar com.twitter.uth.Job --topUsers 10

val raw = args.optional("topUsers")
raw.foreach(s => require(Try(s.trim.toInt).isSuccess, s"not an int: $s"))

def isIntArg(s: String): Boolean = Try(s.trim.toInt).isSuccess

try preferredInt(args, "n", "num", 10) catch { case _: IllegalArgumentException => 10 }

When it happens

Trigger: Running a scalding job with the preferred flag (or its deprecated alias) set to a non-numeric value like '10x' or '100k'; passing a value with stray characters from shell quoting; a scheduled job template with an unreplaced placeholder.

Common situations: Migrating job flags: value supplied to the old deprecated name in a bad format; shell quoting injecting extra characters; cron templates with placeholder text not replaced.

Related errors

• Argument [--entity-type] must be provided. Supported options
• Input user_states $str is invalid. Valid states are: " + Use
• Must read at least some data!
• Adhoc runs require a window: pass --date START END or --mont
• (ScoreType, Entity) ($scoreType, ${entity.toString}) not sup

AI-assisted analysis of xai-org/x-algorithm@24c60942c5 (2026-08-28). Data as JSON: /api/errors/6c1b37a2f3591173. Report an issue: GitHub.