in reply to Re^14: PDL and srand puzzle
in thread PDL and srand puzzle

Recent PDLs, given too few offsets to at (whether because of a syntax mistake or otherwise), throws an error. It allows "too many" offsets, so long as they're all 0, in line with the PDL convention of being able to pretend an ndarray has infinite higher dims all length 1. Your snippet here creates a zero-dim scalar, which then can be either correctly offsetted with no offsets, or over-offsetted with 1 offset.