MathGroup Archive 2010

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

Search the Archive

Tempvar zombies littering my context!

  • To: mathgroup at smc.vnet.net
  • Subject: [mg114424] Tempvar zombies littering my context!
  • From: kj <no.email at please.post>
  • Date: Sat, 4 Dec 2010 06:12:21 -0500 (EST)

I discovered that defining a Module in a certain way causes results
in lingering temporary variable "zombies".


To best show this, first I'll present a simple, zombie-free module:

In[1]:= a[b_] := Module[{c, d},
   c[___] := (Print[c];);
   d[___] := (Print[d];);
   c[];
   d[]
   ]


This works as expected:

In[2]:= a[1];
During evaluation of In[2]:= c$6797
During evaluation of In[2]:= d$6797


Also, the names in the context are as expected; in particular,
there are no temporaries in it:

In[3]:= Names[$Context <> "*"]
Out[3]= {"a", "b", "c", "d"}


The above is what I'd consider "standard operating procedure."

Now, before proceeding to the weirdness, I clear the context, for
good measure:

In[4]:= Scan[Remove, Names[$Context <> "*"]]


...and define another module, identical to the first one, except that
in this one I've added a condition to the last line:

In[5]:= a[b_] := Module[{c, d},
   c[___] := (Print[c];);
   d[___] := (Print[d];);
   c[];
   d[] /; b === 1
   ]


The module's behavior is, again, as expected:

In[6]:= a[1];
During evaluation of In[6]:= c$6798
During evaluation of In[6]:= d$6798


The weirdness is in the context's names:

In[7]:= Names[$Context <> "*"]
Out[7]= {"a", "b", "c", "c$", "d", "d$", "d$6798"}


Now we have lingering temporary variable zombies!  Not only
"stubs" (is this what they're called) like c$ and d$, but also a
full-blown d$6798.  Here are their definitions:

In[8]:= {#, ToString[Definition[#]]} & /@ 
  Select[%, MemberQ[Attributes[#], Temporary] &] // ColumnForm

Out[8]= {"c$", "Attributes[c$] = {Temporary}"},
        {"d$", "Attributes[d$] = {Temporary}"},
        {"d$6798", "Attributes[d$6798] = {Temporary}\n \nd$6798[___] := (Print[d$6798]; )"}}]


FWIW, the zombies appear only when the condition in the last line
is True (leading to the evaluation of the rest of the last line):

In[9]:= a[2];
During evaluation of In[9]:= c$6799

In[10]:= Names[$Context <> "*"]
Out[10]= {"a", "b", "c", "c$", "d", "d$", "d$6798"}


Note that after evaluating a[2] we see no zombie for d$6799.

Nevertheless, every time the condition is true, the context gets
littered with one more zombie:

n[11]:= a[1];
During evaluation of In[11]:= c$6800
During evaluation of In[11]:= d$6800

In[12]:= Names[$Context <> "*"]
Out[12]= {"a", "b", "c", "c$", "d", "d$", "d$6798", "d$6800"}


What's going on?  And, more importantly, how can I redefine the
module in In[5] so that its behavior remains unchanged, but it does
not litter the context with a trail of zombies?

Thanks!

~kj


  • Prev by Date: Re: Replacement Rule with Sqrt in denominator
  • Next by Date: Re: Using a Correlation Matrix to reduce risk
  • Previous by thread: Re: I look for a bi-modal distribution
  • Next by thread: Re: Tempvar zombies littering my context!