Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
1991
*January
*February
*March
*August
*September
*October
*November
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 1991

[Date Index] [Thread Index] [Author Index]

Search the Archive

side effects from setting variables

  • To: mathgroup at yoda.ncsa.uiuc.edu
  • Subject: side effects from setting variables
  • From: CAMERON at midd.cc.middlebury.edu
  • Date: Fri, 22 Feb 91 17:17 EST

I have thought of a potentially evil trick.  I don't have a use
for it as yet, but I could think of a few.  It came to me suddenly,
and it was so delightfully perverse that I wanted to share it.

The thought that led to this was the realization that in many
systems (including Mma) there are "magic variables" that cause
things to happen "behind the scenes" when their values are set.
This example shows that it's easy to make Mma do an arbitrary side
computation when setting a variable, and at the same time simulate
the setting of the variable so that querying the variable returns
what you'd expect.

Note: in the definition of "foo", the expressions "bar = bop = x" and
"arbitrary[x]" are just examples to show what can be done, but the
last two expressions "foo := x" and "x" are necessary -- and they
must be last, and in that order -- to make the "assignment" to "foo"
returned the assigned value at the time of the assignment and in future
evaluations of "foo".

Mathematica (NeXT) 1.2 (January 20, 1990) [With pre-loaded data]
by S. Wolfram, D. Grayson, R. Maeder, H. Cejtin,
   S. Omohundro, D. Ballman and J. Keiper
with I. Rivin and D. Withoff
Copyright 1988,1989,1990 Wolfram Research Inc.

In[1]:= arbitrary[x_] := Print["An arbitrary function has been called on ",x]

In[2]:= foo/: Set[foo,x_] := (bar = bop = x; arbitrary[x]; foo := x; x)

In[3]:= {foo,bar,bop}  (* initially, these have no value -- including foo! *)

Out[3]= {foo, bar, bop}

In[4]:= foo = 44 (* now the fun begins! *)
An arbitrary function has been called on 44

Out[4]= 44

In[5]:= {foo, bar, bop}

Out[5]= {44, 44, 44}

In[6]:= ?foo
foo
foo/: foo := 44

foo/: (foo = x_) := (bar = bop = x; arbitrary[x]; foo := x; x)


In[6]:= ?bar
bar
bar = 44

In[6]:= foo = 83 (* if we "reset" foo, the right things happen *)
An arbitrary function has been called on 83

Out[6]= 83

In[7]:= {foo,bar,bop}

Out[7]= {83, 83, 83}

In[8]:= ?foo
foo
foo/: foo := 83

foo/: (foo = x_) := (bar = bop = x; arbitrary[x]; foo := x; x)


In[8]:= ?bop
bop
bop = 83

In[8]:= t = 99

Out[8]= 99

In[9]:= foo = t^2 + 2
An arbitrary function has been called on 9803

Out[9]= 9803

In[10]:= ?foo
foo
foo/: foo := 9803

foo/: (foo = x_) := (bar = bop = x; arbitrary[x]; foo := x; x)


In[10]:= ?bar
bar
bar = 9803

In[10]:= Quit

That last assignment ("foo = t^2 + 2") was a test to verify that
the evaluated result "9803" and not the unevaluated form "t^2 + 2"
would be stored as the "value" of "foo".

--Cameron Smith
  Mathematica maven and free-lance consultant
  cameron at midd.cc.middlebury.edu


  • Prev by Date: Non-Negative Least Squares (Gaussian)
  • Next by Date: Re: data structure equivalent ops
  • Previous by thread: Non-Negative Least Squares (Gaussian)
  • Next by thread: psfix for Mathematica 2.0