J3/01-329 Date: 21 Aug 2001 To: J3 From: Richard Maine Subject: Minloc/maxloc There were some complaints recent on clf about the processor-dependent behavior of minloc and maxloc for zero-sized arrays (and for masks that are all false). Is there any reason why we couldn't define this result? A value of all zeros would seem an obvious choice. Note the similarity to index, which returns zero in comparable situations. Edits: [308:14], [308:22-23], [311:18], [311:26-26] "the value of the result is processor dependent" -> "all elements of the result are zero".