Re: trouble with bind. (fwd)

[Haobo Yu <haoboy@isi.edu>]

That's exactly why there is second constructor in ranvar expo. When you
create it inside c++ you are not supposed to use the constructur with the
bind(); that is only supposed to be used when created from otcl. 

BTW if you are using current snapshot I don't see this problem there.

On Wed, 3 Nov 1999, Zhenhai Duan wrote:

> The message "http://www-mash.cs.berkeley.edu/dist/archive/ns-users/9808/"
> (with the same title) explains the problem.
> 
> There is something waird if (c++) class A contains a class B varible which
> has a bind inside. Please read the expoo.cc code and simply add in another
> ExpoentialRandomVariable variable. Without initialization value, you will
> note the problem.
> 
> --Zhenhai
> --------------------------------------------------------------
> Zhenhai Duan		        PhD student	
> Computer Science Department	http://www.cs.umn.edu/~duan
> University of Minnesota, TC	Phone: (612)626-7526(O)
> -------------------------------------------------------------- 
> 
> On Wed, 3 Nov 1999, Haobo Yu wrote:
> 
> > > This error is different from what you mentioned. And actually I noticed
> > > the reason that the two original ExponentialRandomVariables work is they
> > > are initialized with some values, this makes the contructor to be called
> > > is not the default ExponentialRandomVariable() (which contains a bind
> > > command). 
> > 
> > I don't understand. The constructur with an argument is never called if
> > you allocate it from otcl. If you new it from otcl, bind() will be called
> > and if it's not initialized you'll get a warning. 
> > 
> > And Application/.... is not a random variable, it's a traffic source. I
> > guess you've added avg_ to it but did not initialize. This is irrelevant
> > to ranvar Exponential.
> > 
> > > --Zhenhai 
> > > --------------------------------------------------------------
> > > Zhenhai Duan		        PhD student	
> > > Computer Science Department	http://www.cs.umn.edu/~duan
> > > University of Minnesota, TC	Phone: (612)626-7526(O)
> > > -------------------------------------------------------------- 
> > > 
> > > On Wed, 3 Nov 1999, Haobo Yu wrote:
> > > 
> > > > when you adds a new var you _should_ initialize it in ns-defaults.tcl
> > > > otherwise this warning results.
> > > > 
> > > > > > ---------- Forwarded message ----------
> > > > > > Date: Tue, 2 Nov 1999 22:17:39 -0600 (CST)
> > > > > > From: Zhenhai Duan <duan@cs.umn.edu>
> > > > > > To: Ns Mailing List <ns-users@mash.CS.Berkeley.EDU>
> > > > > > Subject: trouble with bind.
> > > > > > 
> > > > > > In the message
> > > > > > "http://www-mash.cs.berkeley.edu/dist/archive/ns-users/9808/", Nitin
> > > > > > Vaidya mentioned one problem with bind. I met this problem when I modified
> > > > > > the code of expoo.cc to add another ExponentialRandomVariable variable.
> > > > > > And got the following message:
> > > > > > 
> > > > > > warning: no class variable Application/Traffic/Exponential::avg_
> > > > > > 
> > > > > >         see tcl-object.tcl in tclcl for info about this warning.
> > > > > > 
> > > > > > I am curious that why the original two ExponentialRandomVariables work
> > > > > > without warning.
> > > > > > 
> > > > > > Thanks.
> > > > > > --Zhenhai
> > > > > > --------------------------------------------------------------
> > > > > > Zhenhai Duan		        PhD student	
> > > > > > Computer Science Department	http://www.cs.umn.edu/~duan
> > > > > > University of Minnesota, TC	Phone: (612)626-7526(O)
> > > > > > -------------------------------------------------------------- 
> > > > > > 
> > > > > > 
> > > > > > 
> > > > > 
> > > > > 
> > > > 
> > > > 
> > > 
> > > 
> > 
> 
>