It's not a tolerant comparison (not that it matters for comparing 0 with 0!). If it was, they would use the standardese "if A is tolerantly-equal-to B". Also section 12.2.1 has an informal note listing functions which use comparison-tolerance, and the list does not include Logarithm.
Jay. On 13 July 2016 at 12:21, Juergen Sauermann <juergen.sauerm...@t-online.de> wrote: > Hi Jay, > > I see. Which leaves the question if "equal" shall be strict or within > *⎕IO*. > > Since we are dealing with real numbers (and therefore often rounding > errors) within *⎕IO* > makes more sense to me but the standard does not mention *⎕IO* for *⍟*. > > /// Jürgen > > > On 07/13/2016 01:05 PM, Jay Foad wrote: > > My ISO ("First edition 2001-02-01") says: > > Evaluation Sequence: > If either of A or B are not numbers signal domain-error. > If A and B are equal, return one. > If A is one, signal domain-error. > Set A1 to the natural-logarithm of A. > Set B1 to the natural-logarithm of B. > Return B1 divided-by A1. > > 0⍟0 falls into the "A and B are equal" case. > > Jay. > > On 13 July 2016 at 11:04, Juergen Sauermann <juergen.sauerm...@t-online.de > > wrote: > >> Hi Kacper, >> >> my ISO (June 9, 2000) says DOMAIN ERROR. >> >> More precisely, they say that A⍟B is (⍟A)÷(⍟B) and then that ⍟0 gives >> DOMAIN ERROR. >> >> /// Jürgen >> >> >> On 07/12/2016 11:43 PM, Kacper Gutowski wrote: >> >> According to ISO, 0⍟0 should be one. >> GNU APL gives: >> >> 0⍟0 >> DOMAIN ERROR >> 0⍟0 >> ^^ >> >> -k >> >> >> >> >> > >