🔒 Locked

Struggling with maps in C++

Started by Nauraushaun Jan 17, 2011 at 12:14 AM 7 replies 1.6k views

Original Post

Jan 17, 2011 at 12:14 AM

I've got a map that contains char pointers, and when I try to access it in certain ways it doesn't let me, it just creates a new object that's identical to the one that's already there. Very frustrating. Here's a code sample:

int nBuffer = m_mKeyValues[uKeyNameBuffer]; // nBuffer == 0, the item is not found
char* uKeyNameBuffer2 = "Q";
int nBuffer = m_mKeyValues[uKeyNameBuffer2]; // nBuffer == 16, the item is found

In the first instance, using breakpoints and debugging I've deciphered that uKeyNameBuffer contains "Q". But it won't find the "Q" item that exists in the map, it instead creates one with the exact same key (which should be an impossibility as far as I can tell). The second variable, uKeyNameBuffer2, when checked with breakpoints contains the exact same data in every way as uKeyNameBuffer...so why is the map treating it differently?

So far as I can tell it's a glitch in the map class. The entire point of each item having a key is that you access it with that key, and you can't create another item with the same key. But that's exactly what it's forcing me to do. Of course, it's probably just my stupidity getting in the way, and it's extremely unlikely that the map class would have such a large flaw...but what on earth is going on?

nobodynews

Jan 17, 2011 at 12:21 AM

I've got a map that contains char pointers

I'm not going to bother reading the rest of your post and just tell you to use std::string instead of char *. There's a good chance your problem is due to std::map not comparing the strings but the addresses of the char *s. Use std::string and if the problem persists then get back to us.

PrestoChung

Jan 17, 2011 at 12:25 AM

A char* is an address, no? It's probably not looking further than that.

Nauraushaun

Jan 17, 2011 at 12:31 AM

That'd be it, it's got a different address...But then why does it have the same address as the version in the map if I use the "Q" method? Using debugging I've found that it does indeed have the same address as the item in the map, but how can this happen? I choose to store a Q in a variable, and it automatically makes it the same address as another Q? Is it some kind of optimization?

nobodynews

Jan 17, 2011 at 12:32 AM

I hesitate to bring this up because I think the best solution is to use std::string, but in the interest of full-disclosure there is a way to use std::map and c-strings. Example here:

http://www.sgi.com/tech/stl/Map.html

Juanxo

Jan 17, 2011 at 1:39 AM

Is it some kind of optimization? [/quote]

yes, some compilers store const char* in the same address as an optimization

iMalc

Jan 17, 2011 at 6:18 AM

Is it some kind of optimization?

yes, some compilers store const char* in the same address as an optimization
[/quote]Absolutely, and some compilers go even further than that. An app with strings "Hello world" and "world" might result in pointers to the string "world" pointing about six characters into the middle of "Hello world".

Nauraushaun

Jan 20, 2011 at 10:58 PM

Sorry I didn't post in a while, but I got through it by comparing using strcmp, rather than the equality operator.
Thanks everyone for all your help.

Zao

Jan 21, 2011 at 8:28 AM

As some may have hinted to above, the default comparator for a [font="Courier New"]std::map[/font] is [font="Courier New"]std::less[/font], which defaults to [font="Courier New"]operator <[/font], which for pointers just compares addresses.
As mentioned, string literals may be folded together in static storage, ending up comparing equal by coincidence.

The proper solution here if you really must use [font="Courier New"]char const*[/font] as your key is to provide a comparator to your [font="Courier New"]std::map[/font], which would do a string comparison with [font="Courier New"]strcmp[/font] or a similar function.
Of course, by using raw pointers you run into ownership issues, as you have no way of properly handling the memory the pointers point to. Changing the key type to [font="Courier New"]std::string[/font] would most probably be a Good Thing.

Note that [font="Courier New"]operator [][/font] on a [font="Courier New"]std::map[/font] default-constructs a value if the key could not be found and inserts it into the container. If you do not desire that behavior, use the [font="Courier New"]find(key)[/font] member function which returns an iterator to a key-value pair if the key was found, or the [font="Courier New"]end()[/font] iterator if it was not found.

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.