stl nested maps

I have put some kind of pseudocode for the issue I am facing

I have a hybrid map, I want to insert a particular element from this map to another map.

typedef std::map<int,int> MapOfInts;
typedef std::map<int,MapOfInts> HybridMapOfInts;

Code:

---

int main ()
{
  MapOfInts mymap;
  MapOfInts mylatmap;
  HybridMapOfInts myNewmap;
  MapOfInts::iterator it;
  HybridMapOfInts::iterator itNew;

  mymap[5]=50;

  mylatmap[1]=10;
  mylatmap[2]=20;

  myNewmap.insert(std::make_pair(1, mylatmap) ); 

  itNew = myNewmap.find(1);

//Following gives error, is there any efficient simple right way to do this ??
  mymap.insert(itNew->second);
}

---

Last line gives error , is there any efficient simple right way to do this ??

You have a type issue here.

- itNew is an iterator to a pair<int, MapOfInts>.
- myMap IS a MapOfInts.

What exactly do you want "mymap.insert(itNew);" to do?

Thanks a lot Monarch for the inputs.

Yes sorry, but my last statemaent should have been something like:
mymap.insert(itNew->second);

I want to add (append) the element itNew from the mylatmap to the the mymap.

#####
Basically scenario is, I have final "mymap". I am sending some info in the Request message ( with some Identity) and storing the "data" which is similar to my map in temp storage. Once response is received, I retrive the data from temp storage and populate in final "mymap". So i use the "Identity" as key and get the value which is populated into the final map.

Please try to provide actual code that you have written for real. These little inaccuracies make helping you difficult.

Anyways, map has an insert variant that takes an iterator range, so you can simply do:

Code:

---

mymap.insert((itNew->second).begin(), (itNew->second).end());

---

It's as efficient as inserting a container (non-destructivelly) into another can get, which is to say: o(n).

Quote:

---

Originally Posted by monarch_dodra

It's as efficient as inserting a container (non-destructivelly) into another can get, which is to say: o(n).

---

Not quite: there may be as many insertions as there are objects in the range, but each insertion would normnally take logarithmic time.

Quote:

---

Originally Posted by laserlight

Not quite: there may be as many insertions as there are objects in the range, but each insertion would normnally take logarithmic time.

---

Yes, very true. An oversight on my part. I just meant to say that as a general rule, you shouldn't hope for better than o(N). (At least, in a non-destructive fashion, and containers that don't share their internals, AFAIK).

That said, I was looking at the standards.
C++98 has this to say: "Nlog(size()+N) (N is the distance from i to j) in general; linear if [i, j) is sorted according to value_comp()"
but C++11 has this relaxed to simply: "N log(a.size() + N) (N has the value distance(i, j)"

Although cplusplus.com does state: "Implementations may optimize if the range is already sorted."

The more you know I guess. It's interesting to discover this little changes from 98 to 11.

@monarch: Thanks a lot for all the inputs. Sorry for the delay.

Btw now i have provided some parts of my actual code

Code:

---

CPCRF.h:
struct PolicyRuleInfo{
        BearerQoSInfo stBearerQoS;
        TFTInfo stTFTInfo;

        PolicyRuleInfo(){};
        PolicyRuleInfo( BearerQoSInfo const& qos, TFTInfo const& tft)
      : stBearerQoS(qos), stTFTInfo(tft)
    { }
};

typedef map<string, PolicyRuleInfo> listOfPolicyRuleInfo;

struct IPAddressPolicyRulesInfo{
        CIPAddress sIpAddress;
        listOfPolicyRuleInfo policyRules;
        IPAddressPolicyRulesInfo(){};
        IPAddressPolicyRulesInfo(CIPAddress ipaddr, string policyRuleName, PolicyRuleInfo policyRule): sIpAddress(ipaddr){policyRules[policyRuleName]=policyRule;};

        void addPolicycyRule(string policyRuleName, PolicyRuleInfo policyRule) { policyRules[policyRuleName]=policyRule; }
};

// Apn to PolicyRules
typedef map<string, IPAddressPolicyRulesInfo> APN2PolicyRules;

// Imsi to APN2PolicyRules Info
typedef map<string, APN2PolicyRules>        IMSI2APNPolicyRules;

// GxSessionId to IMSI2APNPolicyRules Info
typedef map<string, IMSI2APNPolicyRules>        GxId2IMSI2APNPolicyRules;

class CPCRF : public CBaseClass
{
public:
:
:
        CPCRF();

        // list of PolicyRules configured for the User
    // ipAddress to User info
        static IMSI2APNPolicyRules        m_mIMSI2ApnPolicyRuleInfo;

        static GxId2IMSI2APNPolicyRules m_mGxId2IMSI2APNPolicyRules;
}

CPCRF.cpp
CompConf::Oam2::IMessagePtr CPCRF::OnAddRequest(CompConf::Oam2::IRequestPtr const& req)
{
        PolicyRuleInfo sPolicyRule(stQoSInfo, stTFT);

        sIpAddPolicyRule.addPolicycyRule(sApn, sPolicyRule);

        sApn2PolicyRule.insert(std::make_pair(sApn, sIpAddPolicyRule));

        IMSI2APNPolicyRules sImsi2Apn;

        sImsi2Apn.insert(std::make_pair(sIMSI, sApn2PolicyRule));

        string sGxsessionId;

        vector<string> stVectTFT;

        stVectTFT.push_back(sTft);

        string sUeIpAddress;

       
        GxRARequest(sGxsessionId, sImsi, sDedBearerId, sUeIpAddress, true, stVectTFT, stQoSInfo);

      // Add this info in the temp map
        m_mGxId2IMSI2APNPolicyRules.insert(std::make_pair(sGxsessionId, sImsi2Apn));
}

CGxDia.cpp                       
DiaRes CGxDiaSvr::RAAnswer(CDiaMessage & sMsg)
{
::
        if(iTerGxId != CPCRF::m_mGxId2IMSI2APNPolicyRules.end())
        {
                //copy this to actual map
        CPCRF::m_mIMSI2ApnPolicyRuleInfo.insert();
                       
            //delete this info in the temp map
                CPCRF::m_mGxId2IMSI2APNPolicyRules.erase(sSessionId);
        }
}

---

Quote:

---

Originally Posted by monarch_dodra

It's interesting to discover this little changes from 98 to 11.

---

here is the corresponding library defect issue

anyway, note that O(n) "sorted insertions" are still guaranteed at container construction...