> Date: Wed, 08 Jan 2003 12:14:10 +0800 > From: Damian Conway <[EMAIL PROTECTED]> > > Can I suggest that an alternative solution might be the following: > > Suppose Perl 6 had two new very low precedence operators: ~> and <~ > (a.k.a. "bind rightwards" and "bind leftwards") > > Suppose ~> takes its left argument and binds it to > the end of the argument list of its right argument, > then evaluates that right argument and returns the result. > So an L2R array-processing chain is: > > @out = @a ~> grep {...} ~> map {...} ~> sort; > > There might also be a be special rule that, if the RHS is > a variable, the LHS is simply assigned to it. Allowing: > > @a ~> grep {...} ~> map {...} ~> sort ~> @a; > > Further suppose that <~ takes its right argument, and binds > it in the indirect object slot of the left argument, which > argument it then calls. So an R2L array-processing chain is: > > @out = sort <~ map {...} <~ grep {...} <~ @a; > > Or, under a special rule for variables on the LHS: > > @out <~ sort <~ map {...} <~ grep {...} <~ @a; > > That way, everything is still a method call, the ultra-low precedence of > <~ and ~> eliminate the need for parens, and (best of all) the expressions > actually *look* like processing sequences.
I think this is a big step towards readability. It allows you to put whatever part of the expression wherever you want (reminiscent of Latin); i.e. always keep the important parts standing out. I also think that the operator (especially a cool 3d-looking one like ~>) is also much more readable than a word in this case. It's a shame <~ is ambiguous. It's a lexical ambiguity, which can be solved with whitespace.... Luke