|
1 - No, maps are hash tables. So, I believe that they generate a hash and use that to decide what bucket to put stuff into
2 - I don't believe so
3 - sure. If you do, it's the memory address that will be hashed, I would think, and I would think that would be as close as you'd get to a random insert order, although I still don't think you can shuffle it.
Christian Graus - Microsoft MVP - C++
|
|
|
|
|
Ok then what's the best data structure for:
1) Being able do a random_shuffle () on,
2) Being able to search by key to get to a value.
I'm using vector right now, but I have a key-value pair combination that I need to update fairly often, so it seems like a map would be ideal for this..., but if this is the case, I need a way of randomly shuffling the map somehow, so when I do an interator and go from the beginning to the end, its random every time.
|
|
|
|
|
Can't you turn the keys into a list and then shuffle that ? How often do you need to shuffle them ? I'd think a list of the keys which you shuffle would be the way to go here.
Christian Graus - Microsoft MVP - C++
|
|
|
|
|
You can get the best of both worlds by combining the two: use a std::map for fast access to data but a std::vector of std::map::iterators into the std::map which you can shuffle. Here's an example using sets.
------------
// Console.cpp : Defines the entry point for the console application.
//
#include "stdafx.h"
#include <iostream>
#include <string>
#include <set>
#include <vector>
#include <algorithm>
#include <iterator>
using namespace std;
typedef std::set<string> StringSet;
typedef StringSet::const_iterator StringSet_CI;
typedef std::vector<StringSet_CI> StringSetIterators;
typedef StringSetIterators::const_iterator StringSetIterators_CI;
int main(int argc, char* argv[])
{
const char* Names[] = {"Bob", "Sue", "James", "Ralph", "Steve", "Paul", "Kim", "Mary"};
const char** pEndNames = &Names[sizeof(Names)/sizeof(Names[0])];
// A map of strings.
StringSet strings;
// Fill the set with the names.
copy(Names, pEndNames, inserter(strings, strings.end()));
// Output the set to the console.
cout << "Names in set:" << endl;
copy(strings.begin(), strings.end(), ostream_iterator<string>(cout, "\n"));
cout << endl << endl;
// Vector of set iterators.
StringSetIterators ssi;
StringSet_CI e = strings.end();
for ( StringSet_CI i=strings.begin(); i!=e; ++i )
{
ssi.push_back(i);
}
for ( int times=1; times<=3; ++times )
{
// Shuffle the iterators.
random_shuffle(ssi.begin(), ssi.end());
// Output through the shuffled iterators.
cout << "Shuffled names:" << endl;
StringSetIterators_CI ee = ssi.end();
for ( StringSetIterators_CI ii=ssi.begin(); ii!=ee; ++ii )
{
cout << *(*ii) << endl;
}
cout << endl << endl;
}
return 0;
}
Steve
|
|
|
|
|
Christian Graus wrote: 1 - No, maps are hash tables.
Actually in general std::maps are red-black trees.
Steve
|
|
|
|
|
Damn - of course they are. I've been in the C# wilderness for far too long...
Christian Graus - Microsoft MVP - C++
|
|
|
|
|
Christian Graus wrote: C# wilderness
So you started as a C++ programmer? How do you find C#? I'm a C++ programmer who is resistant to the whole dotNET phenomenon and as such interested in what a C++ programmer has to say about it.
Steve
|
|
|
|
|
Yes, I was a C++ programmer for a long time. When C# was announced, we made fun of it. Then I moved to writing web apps, and ASP.NET was just streets ahead of ASP, hence the move to C#. I think C# is nice, I like it a lot. I think with 2.0, it's finally turning into it's own language, all the more so when 3.0 comes out, with LINQ, etc. However, overall there are definatley things about it that still bug me to this day. I've fairly recently been doing some C++ work, which has just gone full time, and I'm enjoying it very much. Having said that, I write an image processing app in C#, and that's still happening in my spare time. The focus has moved to C++ again, but I am happy to move between the two languages on a daily basis, they both have strengths and weaknesses.
For a desktop app, I'd say C# is faster, and better supported. C++ is still more powerful and has the better std library. It's a question for me now of choosing the tool for the job.
It still hurts to admit that when I use C#, I'm compiling to the same runtime as VB does now 
Christian Graus - Microsoft MVP - C++
|
|
|
|
|
When you say, "C# is faster" do you mean faster to program or faster at runtime?
Steve
|
|
|
|
|
I took it to mean "faster to develop", but it all depends on your "implementation".
You can easily write some C++ code that doesn't perform well, as well as some C# code that you think will perform well, but doesn't at all, because it's doing so much for you behind the scenes.
- S
50 cups of coffee and you know it's on!
|
|
|
|
|
I accept that if you're going to compare you can't compare a fast algorithm in langauge "a" with a slow one in language "b" and conclude that language "a" is faster.
Steve
|
|
|
|
|
Is there something faster in C# than C++, besides development time? Haven't seen any benchmarks, but I'll bet (something insignificant) C++ will equal or better C#. Bring on the humble pie, I'm hungry... 
(I mean, same algorithm-wise, like quick sort, etc....)
- S
50 cups of coffee and you know it's on!
-- modified at 1:57 Thursday 27th April, 2006
|
|
|
|
|
Steve Echols wrote: Is there something faster in C# than C++
I would also say no. But it's one thing so guess and another to know. Also, even if it is slower at runtime, one could still ask "how much slower?" - Did the lack of runtime speed present a problem?
Steve
|
|
|
|
|
Yeah, I don't know either, but I would guess no. I've only done a few apps in C#, and the speed of something equivalent in C++, on the same machine, was very disappointing.
It's not just the runtime load that bothers me, it's the speed of the whole app that seems "sluggish". I thought once it was JITted, then everything is cool, but it seems like every .net app I've used is just painfully slow! (Maybe I should upgrade my 386 )
- S
50 cups of coffee and you know it's on!
|
|
|
|
|
C# is faster to develop in, especially GUI stuff. C++ is plainly faster assuming that both code sets impliment the same algorithms. I did move code from C# to C++ once before tho, and it was a LOT slower. When I was done, it was a fair bit faster, I forget why it was so slow to start with, something to do with passing parameters forcing a copy, I think.
Christian Graus - Microsoft MVP - C++
|
|
|
|
|
Is there anyway of shuffling a map, or other hashtable like thing?
For instance, most of the time I want to iterate from the first to last element... but about once an hour, I need to be able to select an element based on a key and update its value.
On the hour I am taking data from a list of m elements and updating n elements (where n>m, and n < 100). And there is a key that elements m and n share in common.
Do I optimize for the general case which is just general cycling thru a list from beginning to end? I must also shuffle every time. So I was thinking a vector would be ideal.
Your thoughts?
Thanks!
|
|
|
|
|
Yeah, I think a vector sounds like the best choice.
Steve
|
|
|
|
|
HI.
I have alwasy assumed for some bad reason that DBL_MIN means the minimum value for a double. But it is actually the minimum position number.
I was wondering is there a CONSTANT that holds the minimum double (negative) number?
Thanks in advance.
|
|
|
|
|
First include <limits>. Now get minimum like this:
-std::numeric_limits<double>::max();</code>
Steve
|
|
|
|
|
DO I also need to do something else? It says that std is not a namesapce etc. THanks for the help.
|
|
|
|
|
Just have to make sure you have this line in the file:
#include <limits>
Steve
|
|
|
|
|
For strict ANSI you should only use <limits.h> and assume the the most negative double representable is -DBL_MAX (the negative of DBL_MAX). This will result in -1.7976931348623158e+308, which is, in fact the most negative number that can be represented.
This assumption is reasonable since the sign bit can always be used to represent any negative form of any positive number and vice-versa.
Note, however, that the "resolution" is only of about 15 digits. This means that 10e20 + 1.0 will result in 10e20. And you can keep adding 1.0 and the result will never change from 10e20.
So, it is important to note the difference between dynamic range (308 exponent) and resolution (about 15 decimal digits).
Rilhas
|
|
|
|
|
Hi.
I’m trying to develop an application that operates on the oracle database using (ODBC) CDatabase & CRecordset. But the application can't show the record (NUMBER) type correctly and it show instead some strange numbers in the edit contrl. -19841 something like this.
// here's an example of the code.
char str[255];
int index=0;
CDBVariant var;
record.GetFieldValue(index,var);
wsprintf(str,"%d",var.m_iVal);
EditCtrl->SetWindowText(str);
Somebody tell me please what I have to do to convert the number and show it correctly? 
I'm using VC++ 6.0 & Oracle 9i (9.2.0.1.0)
Best regards. 
|
|
|
|
|
M_Nuaimi wrote: char str[255];
int index=0;
CDBVariant var;
record.GetFieldValue(index,var);
wsprintf(str,"%d",var.m_iVal);
EditCtrl->SetWindowText(str);
You are using char with with wsprintf! Could be a reason. Please check.
Nibu thomas
Software Developer
|
|
|
|
|
I would have thought that would have generated a compiler error....
Steve
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
