[667] in Coldmud discussion meeting

Re: assignment operator

daemon@ATHENA.MIT.EDU (Tue Mar 7 16:22:52 1995 )

From: brandon@petersboro.declab.usu.edu
To: coldstuff@MIT.EDU
Date: Tue, 07 Mar 95 14:15:42 -0700

< I'd like to propose '->' be the assignment operator. It visually represents
< what happens to the variable being assigned the new value. It has no chance
< of being pronounced, 'equals'. It looks cool...

Oh, hey, I like it, except for I would use the alternate as it looks more
appropriate, i.e.:

----------
while (success <- .attempt()) {
  .display(success);
}

if (where <- .location_of(foo)) {
  .work_with(where, foo);
} else if (where <- .location_of(bar)) {
  .work_with(where, bar);
} else {
  .not_working();
}
----------

The '->' looks too much like you are (somehow) assigning the value on the left
to the right...

Of course, we do need to consider the standards issue.  In most languages
an assignment (operator, or in any form) usually consists of a character
sequence which includes an equals sign of some sort.  There is also the
consideration that -> is already used in C, and this _is_ a somewhat C-like
language (of course using '=:' isn't much worse).  Too bad '<' with equals
signs is already used...

Actually, I was just thinking that the main drawback to '=:' is that it looks
too much like some convulted "EQUALS x OR x" i.e. "foo=x:y;".  Then I realized
that in ColdC ':' isn't used in that way, so it would not be a confusion...

Hmm, ok, my vote would be for either '=:' or ':=', simply because (even
though it is cool) '->' or '<-' would be too far from normalness...

Some samples:

----------
while (success := .attempt()) {
  .display(success);
}

if (where := .location_of(foo)) {
  .work_with(where, foo);
} else if (where := .location_of(bar)) {
  .work_with(where, bar);
} else {
  .not_working();
}
----------

The other issue I've been beating around is the one initially brought about
by the "local" hack from Jordan.  I've been considering a few ways of
implementing "local" and "global" methods.  The main bump I have is in
method lookups.  It can:

    use the same operator ('.'), which means that for every method it
    would have to search both the global list and local list.  However,
    if you accidentally call a local method globally, it can return
    an intelligent error (rather than "HELLO, WAKEUP, ANYBODY THERE???").

Or:

    use a different operator (??), which means that it would speed
    up lookups (as it knows exactly which list to use), but it would
    include a new operator (operator conflict dillemma arises) and it
    would be less intuitive (i.e. if we used ':' all the MOO-ers would
    be writing entire objects which local only methods 8)

Of course, we could use an odd bent on the C method (of either '.' or
'->'), and have global lookups with '.' and local lookups with '->'
(or perhaps invert it, so local lookups are with '.' and global lookups
are with '->', would be closer to the pointer idea in C (i.e. pointing
to a method off this object).  For instance...:

---------
foo = .foo_method();
.set_foo($parse->parse_foo(foo));
->tell(foo);
---------

Of course, then you would want to use '.' to lookup global methods which
are on this object (this wouldn't be too much of a problem though...)

-Brandon (Lynx)-