I agree about the above books with a few notes:
The OpenGL programming 8th guide is now out and has be redone for modern OpenGL 4.3.
The SuperBible 5th ed, does provide you will a crutch library to start off but as you go through the book you reimplement the functionality of that library so by the end of it you should be fully versed. I also highly recommend wrapping OpenGL stuff in something if your using a object orientated language. I'm working on my own library for fun which seems to be the way to go, implementing it myself I understand all the ins and outs and I only need to put in the bits I actually use. Also you can abstract away some things like sending data to shaders (isn't shinyMaterial.setColor(Red) much nicer), this can also let you backport some stuff by providing legacy alternatives. There is also oglplus.
The main problem with learning modern 'core' OpenGL is that to get a basic pipeline up and running even if it's just to draw a cube you need to understand the following:
- Matrix operations (use GLM if you on C++, you will also need a MatrixStack OpenGL Unofficial SDK's version here, or my simple one here.)
- GLSL shader programming.
- binding data to shaders and Vertex Buffer Objects.
If you wanted to learn just real pure core profile they would all need to be implemented simultaneously and if anything is broken you just wont see anything and have almost no clue where it went wrong. That also doesn't cover things like basic lighting (where you need to learn the shading algorithms in addition to the programming language).
But you don't need to start in the deepend with the core profile. Use some of the deprecated functionality at the beginning, just about everything will still support it, then modernize it.
Another issue is that 'modern' OpenGL can really refer to 3 things:
The first one is the programmable shader pipline as opposed to the fixed function pipeline. The core 'modern' techniques have been possible since OpenGL 2.0 was released in 2004 when shaders where introduced (the tutorials just didn't get updated) also 1.5 had Vertex Buffer Objects's which are another cornerstone of modern OpenGL. A problem is that even today OpenGL developers might not even be able to take advantage of all that stuff, there are plenty of crappy netbooks out there with those Intel chipsets that only support OpenGL 1.5 and phones such as older iPhones/Android ones that only support OpenGL ES 1.1. The only real difference between this stuff in OpenGL 2.0 and OpenGL 3.x/4.x is it's now mandatory for the core profile.
Modern OpenGL could also refer to the new stuff that's in 3.x, 4.x. Things like Vertex Array Objects, Framebuffer Objects, Uniform Buffer Objects/std140, layout qualifiers, GLSL subroutines and so on. Many of those I have found can make things much much nicer to program with. They can also subtly change your overall engine pipelines. They are also things that with a little effort you can backport by making helper functions and so on (For example write a generic shader manager object that you can use to set uniforms, if UBOs are supported use that otherwise manage uploading the values to shaders manually, maybe put a little shader rewriting stuff in there to automate it). This stuff only runs on new video cards, you wont see any netbooks supporting it or any current generation video game consoles (they use their own subset of OpenGL similar to OpenGL ES). The latest OSX only have 3.2 support and I understand that they will be fixed at this version forever since Apple write their graphics drivers and specify OpenGL versions based on the OS version not the GFX drivers/video card capabilities the only way for a user to get a new OpenGL on a current generation Mac would be to pay for an upgrade to a new version of OSX.
The Ultra Modern nextgen OpenGL. OpenGL 4.3 was just released. It comes with compute shaders, glMultiDrawElementsIndirect and shader storage buffers. This allows for a fundamentally different pipeline where you upload all the vertex data to the GPU and then use compute shaders to calculate what should be drawn rather than doing it in your program. Unfortunately this is very new so I don't know of any decent tutorials on it. Take a look at the OpenGL 4.3 review which gives an outline of how this works (also the author says there is a GPU Pro article to be released explaining it). Being brand new it's not going to be supported anywhere at this stage (unless you're on a newer nvidia card with beta drivers). Although you might be able to use OpenCL/CUDA to implement some of it.
I would skip glBegin, glEnd, glVertex*, glNormal*, glTextCoord*, glTranslate*, glRotate*, glLoadIdenity, glModelViewMatrix and so on. Instead start with learning Vertex Buffer Objects without shaders, don't try 3D at this point just a flat white triangle so you don't have to work with 3d operations using matrices (as you will need to learn to bind them to shaders soon so no sense in learning to old way). Next add in in the shaders, you will need to change how your vertex buffer objects are bound since you will now want to use vertex attributes. Learn how to communicate properties to them via uniforms. Then learn matrix operations with GLM, bind the matrices to the shaders and get 3D going. Once you have that done then it's basically just learning more advanced things like texturing, lighting and some special features like framebuffer objects, vertex array objects, uniform buffer objects, GLSL routines and so on (those are all things I recommend you have a look at. Most of it is fairly new except for texturing and lighting but if you learn them as shader concepts rather than as OpenGL ones, you should get the modern approach).
Some general online modern OpenGL tutorials, covering the shader programming pipeline that most of which works with OpenGL 2.0+:
For other general learning resources. OpenGL ES might be a good starting point. OpenGL ES has stripped out all the deprecated stuff. OpenGL 4.1 has full compatibility with OpenGL ES so it should be possible to take an OpenGL ES program and run it in OpenGL 4.1. And the basic principles are the same for both.
Some things to note is that OpenGL ES 3.0 has just been released but as it's new, it won't have any tutorials or hardware support yet. It doesn't seem to have any major fundamental changes (like the difference between fixed pipeline and shader model that we saw with OpenGL ES 1.0->2.0). So you probably won't be learning any out of date stuff.
Also, OpenGL ES will often be using EGL so you will need to learn another way to open a Window but that shouldn't be too hard. Otherwise there are a few OpenGL ES wrappers/emulators around or you can learn on Android/iPhone if you have one (or look at using their emulators).
There is the Addison Wesley OpenGL ES 2.0 Programming Guide which is the equivalent of the red book.
You can also look at GLSL shader focused stuff since that's where most of modern OpenGL lives. You just need to ensure what you learn covers the OpenGL binding 'glue' side of things.
Check out the OpenGL Shading Language 2nd edition (Orange Book).
The OpenGL 4.0 Shading Language Cookbook covers some more bleeding edge OpenGL stuff but isn't itself a comprehensive tutorial. Its definitely worth a look to get some understanding in the latest real world OpenGL practices such as uniform buffer objects and GLSL subroutines.