Discussion: new math operators

[R.M.Everson]

Neil Hodgson <neilh at scintilla.org> wrote:
  >> As one of the more reactionary Pythoneers with regard to syntax
  >> changes and as someone who is unlikely to ever need to use the
  >> new operators, I say the braced operators are the best.

Huaiyu Zhu <hzhu at knowledgetrack.com> writes:
   > I just posted a summary that dismissed this idea before seeing
   > recent two posts favoring brace based operators.  It's becoming
   > increasingly obvious that most discussants here favor this. Yes,
   > I agree with all the advantages of this.

   > Now the questions are:
   > 1. How do we test reactions from people who do not read this
   >    thread?  Anyone volunteer to give a try?

   > 2. How important does the possible future extension play in this?
   >    What considerations should be put in this?

   > 3. If there could be future calculus of operators, would we now
   >    favor (*)=
   > or (*=) for augmented assignment?

I've been following this thread for a while.  As a Matlab user, I am
quite comfortable with the .op notation and could certainly live with
the @op variant, although I prefer ~op as being less dense and
P***-ish.  These have the (distinct) advantage of being quite
straight-forward to implement in the very short term.

However, I certainly agree with the arguments for (op) except in one
respect.  As Huaiyu demonstrated in an earlier post, I find that style
very difficult to parse visually, particularly  anything involving
parentheses to group expressions.  Easy visual parsing is a
prerequisite for this sort of application where there is a good deal
of interactive, exploratory development: it's one of the things that
makes Python (and Matlab) a pleasure.

Most of the confusion arises because () are already over-used (for
grouping, forcing precedence, enclosing tuples, enclosing function
arguments, and probably more).  How about a different pair of
delimiter characters to enclose the operator?  I would rule out {} and
[] as they also have quite specific meanings already.  

That leaves <op>

Two possible objections:
  1.  Ambiguity with greater/less than.  I haven't thought of any
      situations that couldn't be resolved by a decent parser.
  2.  It looks a bit like bra and ket, but isn't.


I would greatly prefer this syntax to (op).

Richard.

-- 
Richard Everson.
Department of Computer Science, Exeter University    Voice: +44 1392 264065
R.M.Everson at ex.ac.uk                                   Sec: +44 1392 264061
http://www.dcs.ex.ac.uk/academics/reverson             Fax: +44 1392 264067