Fonts, don't we all love fonts...

[Maximilian]

yeah... I'm really REALLY sorry for wasting so much time... Not used to working with librariest I didnt write myself.

Ok, here goes:
I'm using MS Visual Studio Express 2012

This is my project file:
https://www.dropbox.com/s/40mkcnxyee0vhe4/GraphClass.vcxproj

And I downloaded the vc10 package. I did this because I'm uncomfortable working with other people libraries (and in my profession I work with entirerly different types of environments, often entierly preconstructed environments where I just ahve to write code...)
I'm Fairly certain I'm using the 32-bit library, for various reasons.

[Laurent]

I'm using MS Visual Studio Express 2012

And I downloaded the vc10 package

What about downloading the package that matches your compiler?

[Maximilian]

Ther IS one for vs 2012? I havent seen one...

EDIT: IM AN IDIOT!
Nevermind me... Im just gonna go and be ashamed of myself now...

[Nexus]

Please trust me when I say this: My code runs a simple graph datastructure. I have spent alot of time on projects like this one before and my errors are not related to missmanagedment of memory on my part. Yet the problem persists.

Even if it's not the problem in this case, you can greatly improve your code with the RAII idiom. There are so many problems with raw memory management, of which memory leaks is only the most obvious one. Since C++ makes it so simple to handle memory automatically (yes, your code will become simpler) and allows to do this at zero cost (no, your code won't be slower), there is virtually no reason to still manage memory by hand.

I tried to show an example of how much can go wrong in a simple code in this thread, please read it. I'm sure you'll like RAII

[FRex]

Maybe it's about time we add a self validation feature so SFML detects/warns it's being used in a different compiler than the one in which it was compiled?
I'm not even joking at this point, self validation is dumb but so are 2 pages of 'WAAFONTBROKEN' that in the end is non interoperable compilers issue, again...

[Laurent]

Maybe it's about time we add a self validation feature so SFML detects/warns it's being used in a different compiler than the one in which it was compiled?

This would be difficult. It's not just the compiler version, it's a lot of configuration flags: standard library linkage, threading model, exception model, C++11, ... basically anything that breaks binary compatibility between two variants of a compiler.
And I don't think this should be checked at runtime by the library anyway. By the way, it doesn't happen that much, the most common mistake is rather to use release libraries in debug mode.