Kealey, Martin, wrote: > Nice idea; introduces one particular ambiguity though: would > > $m ** $n > > then be > > pow($m, $n) > > or > > pow($n, $m) > > ?
Neither. As with the reducing meta-operator, you would need to have the ability to define an operator that takes precedence over a meta'd operator, for those cases where a particular combination of symbols has a meaning that is either more broad or incompatible with what the replication meta would supply. Conveniently, '**' is one such operator. To whit: In '$count *op $value' and '$value op* $count', $count must be an unsigned integer of some sort: it's nonsense to replicate an operation half of a time, or negative two times, or i times. However, '$base ** $power' places no such restrictions on either $base or $power; thus, it needs to be separately defined. The fact that its behavior coincides with a replicator appended to a multiplication operator when the power is an unsigned integer is merely a convenient coincidence as far as the compiler is concerned. That said, there _is_ still some potential ambiguity if you have operators that begin or end with '*'. For instance: let's say that you want to apply replication to exponentials: does '$m *** $n' mean '$m repetitions of $n ** $n', or does it mean '$m ** $m, repeated $n times'? The longest-token rule resolves this dilemma in favor of the latter. -- Jonathan "Dataweaver" Lang
