No this is the wrong way to go about this.
How are you going to do trap detection? What about when the player reaches the edges of the walls? Will your viewing system work for dungeons or will you have to re-write a significant portion of the code?
The world is geometry. The player is geometry. The world doesn't move. The player does. Set the camera's position to center on the player. Always. And that's all there is to it.
Don't try and get fancy with "oh if I slide the world, then it will give the appearance the player is moving". You're just going to complicate the math with weird coordinate systems at the end of the day.
It is true that OpenGL's rendering actually works by "fixing the camera to point down - z, and transforming and rotate all world geometry so it fits in the canonical view volume", but you're not meant to think about it that way when programming.
gluLookAt has parameters named
up for a reason -- so you can think in terms of a sensible coordinate system.