Re: [ns] target_ is (NSObject *)0x0

[Lee <yli@gpu.srv.ualberta.ca>]

Thanks Scooter,

I've figured out the bug...
The agent was initilized by mistake :(

Cheers,
Lee

On Sun, 14 Jan 2001, Scott Michel wrote:

> FWIW, my guess is that there is an intermediate node somewhere
> in your sim that doesn't have an agent attached to the node.
> 
> Also, you might want to dump the agent's dest address and port
> fields to see if they are properly (not randomly) initialized.
> Alternatively, you might also dump the dest of the packet you're
> sending to insure that the packet's dest is legit.
> 
> I hacked config.h to provide a default ctor for ns_addr_t so
> that these would be readily obvious to me when I did something
> totally dippy, like not connecting an agent to a destination.
> (Patch follows at end of message)
> 
> 
> 
> -scooter
> 
> On Sat, Jan 13, 2001 at 11:59:08PM -0700, Lee wrote:
> > 
> > Actually, I print out the values in TCL, as follows,
> > 
> >  _o305 attached to 1 with target _o18
> >  _o306 attached to 2 with target _o24
> >  _o307 attached to 3 with target _o30
> >  _o308 attached to 5 with target _o42
> >  _o309 attached to 4 with target _o36
> > 
> > in the format
> > "$agent attached to [$self id] with target $target_ [$self entry]"
> > (it is in ns-node.tcl add-target)
> > 
> > It says that the target_ is not 0x0, it should be the entry of the node
> > (there is $agent target [$self entry] above)
> > 
> > However, in C++, it becomes 0x0.
> > 
> > On Sat, 13 Jan 2001, Lee wrote:
> > 
> > > 
> > > Hello,
> > > 
> > > I got a Segmentation Fault. target_ of some agent is (NSObject *)0x0.
> > > I guess there is something wrong with attaching the agent.
> > > (However I do use attach-agent to do that.)
> > > Would you like to tell me, generally, 
> > > what is the possible reason for that?
> > > Thanks.
> > > 
> > > Have a good time,
> > > Lee
> 
> Index: config.h
> ===================================================================
> RCS file: /work/repo/scottm/ns/config.h,v
> retrieving revision 1.1.1.2
> retrieving revision 1.10
> diff -u -r1.1.1.2 -r1.10
> --- config.h	2000/11/28 20:44:56	1.1.1.2
> +++ config.h	2001/01/04 19:46:46	1.10
> @@ -54,6 +54,10 @@
>  #include "autoconf.h"
>  #endif
>  
> +/* iostream, for ns_addr_t serialization */
> +
> +#include <iostream>
> +
>  /* after autoconf (and HAVE_INT64) we can pick up tclcl.h */
>  #ifndef stand_alone
>  #ifdef __cplusplus
> @@ -77,16 +108,33 @@
>  	int32_t addr_;
>  	int32_t port_;
>  #ifdef __cplusplus
> +	// It's nice to have ctors in C++, no?
> +
> +	ns_addr_t() : addr_(-1), port_(-1)
> +	{ }
> +
> +	ns_addr_t(nsaddr_t addr, nsaddr_t port) : addr_(addr), port_(port)
> +	{ }
> +
>  	ns_addr_t& operator= (const ns_addr_t& n) {
>  		addr_ = n.addr_;
>  		port_ = n.port_;
>  		return (*this);
>  	}
> -	int operator== (const ns_addr_t& n) {
> +	bool operator== (const ns_addr_t& n) const {
>  		return ((addr_ == n.addr_) && (port_ == n.port_));
>  	}
> +	bool operator!= (const ns_addr_t& n) const {
> +		return !operator==(n);
> +	}
>  #endif // __cplusplus
>  };
> +
> +
> +// Generally useful 
> +
> +istream &operator>>(istream &is, ns_addr_t &addr);
> +ostream &operator<<(ostream &os, const ns_addr_t &addr);
>  
>  // 64-bit integer support
>  #if defined(SIZEOF_LONG) && SIZEOF_LONG >= 8
>