[ndnSIM] Question

Alex Afanasyev alexander.afanasyev at ucla.edu
Thu May 9 19:24:55 PDT 2013


Yes. It would work for either lfu or lru policy (even random, though it doesn't make much sense), there will be just different interpretations what is the first, second,... element.

---
Alex

On May 9, 2013, at 6:48 PM, Aaron <aaronishere at qq.com> wrote:

> Hi, Alex
>  
> Is it possible for ns3::ndn::cs::lfu/lru?
>  
> Aaron
>  
> From: Alex Afanasyev
> Date: 2013-05-10 09:20
> To: Aaron
> CC: ndnsim
> Subject: Re: Question
> Hi Aaron,
> 
> Hmm... Again, in general there is no notion of beginning or ending in PIT.  It is just a complex tree-like structure, where items are stored until  they are explicitly removed (interest gets satisfied or timed out).
> 
> The default implementation of PIT in ndnSIM (ns3::ndn::pit::Persistent) has "persistent" item replace policy, meaning that if there is a limit on number of entires and this limit is reached, new entries will be rejected.
> 
> The behavior similar to what you described is happening with ns3::ndn::pit::Lru, where "item in the beginning" is the oldest PIT entry.  Is it what you're asking?  If yes, then you can "remove" second-old, third-old, or older items like this:
> typename policy_container::iterator item = policy_container::begin ();
> std::advance (item, 2);
> if (item != policy_container::end ())
> {
>    base_.erase (&(*item));
> }
> 
> std::advance operation is not very efficient (it has linear complexity), but if you don't go too far away from the beginning, it should work reasonably well.
> 
> ---
> Alex
> 
> On May 9, 2013, at 6:02 PM, "Aaron" <aaronishere at qq.com> wrote:
> 
>> hi, Alex
>> since the insert function always erase the item in the beginning. if it's possible to delete the first several items something like when A comes, erases the beginning, then B comes,erases the second,C erases the third. Say we only replace the first five item, then when F comes, re-erases the beginning
>> ------------------ 原始邮件 ------------------
>> 发件人: "alexander.afanasyev"<alexander.afanasyev at ucla.edu>
>> 发送时间: 2013年5月10日(星期五) 上午7:42
>> 收件人: "aaronishere"<aaronishere at qq.com>;
>> 抄送: "ndnsim at lists.cs.ucla.edu"<ndnsim at lists.cs.ucla.edu>;
>> 主题: Re: Question
>> Hi Aaron,
>> 
>> It is almost impossible to get a position number of an item, since the PIT data structure is a recursive hash-based container.  Is there any specific use case why you want to get this number?  If you    really need such functionality, the custom PIT policy can track PIT item positions.
>> 
>> To iterate over all elements of PIT you can do the following:
>> 
>> Ptr<ndn::Pit> pit = ?;
>> for (Ptr<pit::Entry> entry = pit->Begin (); entry != pit->End (); entry = pit->Next (entry))
>> {
>> ...
>> }
>> 
>> ---
>> Alex
>> 
>> On May 9, 2013, at 4:35 PM, aaronishere <aaronishere at qq.com> wrote:
>> 
>>> Hi, Alex
>>>  
>>>  Is there some way to get the position number of item in the *-policy.h,  such as the lookuped item is the 5th item in the container?
>>>  
>>> Another is how to  loop of all pit-entry, such as
>>> for ( pitentry::begin, pitentry::end, Next( ) )
>>> {
>>> ......
>>> },
>>> I'm not clear abou it.
>>>  
>>> Thanks!!
>>>  
>>> Aaron
> 
> _______________________________________________
> ndnSIM mailing list
> ndnSIM at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20130509/2d133190/attachment.html>


More information about the ndnSIM mailing list