Re: Must use C++ ? OTcl is also enough ?

[John Heidemann <johnh@ISI.EDU>]

On Fri, 17 Sep 1999 10:03:38 PDT, Haobo Yu wrote: 
>Here is my 1 cent: you use C++ if you want
>- to manipulate packets
>- to do large simulation with memory constraint
>- smaller runtime
>
>- Haobo
>
>(OR you are a C++ fun and hate the hacky ad-hockey otcl.)


Haobo has provided some very good rules of thumb.

I'd add:
	- use OTcl if you can do what you want from existing C++ objects

	- use C++ if you have to change the behavior of an existing
		C++ object

For example, links are OTcl objects that assemble delay, queueing, and
possibly loss modules.  If your experiment can be done with those
pieces, great.  If instead you want do something fancier (a special
queueing dicipline or model of loss), then you'll need a new C++
object.

   -John Heidemann


>
>On Fri, 17 Sep 1999, Frank Chu wrote:
>
>> 
>> Very much appreciate if anyone can answer this
>> question.
>> 
>> I want to do some simulation in which I will need to
>> inherit agents of queue management (including
>> dropping), transmission (including rate control), and
>> traffic aggregation.  The simulation will not be too
>> complex.  I saw many people use C++ in their code even
>> the simulation is not complex.  According to your
>> experiences, apart from the purpose that their code
>> can be included into ns, is there anything (any hidden
>> facts?) require they must use C++ ?  I mean, can I
>> just use OTcl if I don't plan my code included into ns
>> ?
>> 
>>   The reason that I have such a question is that I
>> didn't compiled and installed ns successfully on my
>> Unix, but I got a copy of binary code of ns running ok
>> on my machine, so I think probably I can only use OTcl
>> to write my simulation code.
>> 
>> Best,
>> Frank
>> 
>> __________________________________________________
>> Do You Yahoo!?
>> Bid and sell for free at http://auctions.yahoo.com
>> 
>
>