On Tue, Feb 22, 2005 at 09:07:27PM +0800, Autrijus Tang wrote: : On Tue, Feb 22, 2005 at 12:20:35AM -0800, Larry Wall wrote: : > That seems right to me. Or at least right associative. :-) : > : > In any event, right associativity seems more useful in this case. : : Okay, implemented as such. : : Of the "non-chaining binaries" in S06: : : => but does cmp <=> .. ^.. ..^ ^..^ : : How many are similarily right associative? I was listing everything : as non-associative, but it seems that "but" and "does" is left-assoc?
I think the case for associativity is much weaker for "but" and "does", and the direction not intuitively obvious. So let's force people to parenthesize those by keeping them non-associative, I think. Plus it decreases cognitive load to have a consistent rule. (Except for the =>, which makes me wonder if it's at the wrong precedence now. Maybe it should move down to assignment precedence, and then it would naturally be right-associative.) Larry
