Mesa is an open-source implementation of the OpenGL specification. OpenGL is a programming library for writing interactive 3D applications. See the OpenGL website for more information.
Mesa 6.x supports the OpenGL 1.5 specification.
Yes. Specifically, Mesa serves as the OpenGL core for the open-source DRI drivers for XFree86/X.org. See the DRI website for more information.
There have been other hardware drivers for Mesa over the years (such as the 3Dfx Glide/Voodoo driver, an old S3 driver, etc) but the DRI drivers are the modern ones.
Hardware-accelerated OpenGL implementations are available for most popular operating systems today. Still, Mesa serves at least these purposes:
Stand-alone Mesa is the original incarnation of Mesa. On systems running the X Window System it does all its rendering through the Xlib API:
Alternately, Mesa acts as the core for a number of OpenGL hardware drivers within the DRI (Direct Rendering Infrastructure):
This wasn't easy in the past. Now, the DRI drivers are included in the Mesa tree and can be compiled separately from the X server. Just follow the Mesa compilation instructions.
Yes, SGI's OpenGL Sample Implemenation (SI) is available. The SI was written during the time that OpenGL was originally designed. Unfortunately, development of the SI has stagnated. Mesa is much more up to date with modern features and extensions.
Vincent is an open-source implementation of OpenGL ES for mobile devices.
miniGL is a subset of OpenGL for PalmOS devices.
TinyGL is a subset of OpenGL.
SoftGL is an OpenGL subset for mobile devices.
Chromium isn't a conventional OpenGL implementation (it's layered upon OpenGL), but it does export the OpenGL API. It allows tiled rendering, sort-last rendering, etc.
ClosedGL is an OpenGL subset library for TI graphing calculators.
There may be other open OpenGL implementations, but Mesa is the most popular and feature-complete.
If you're using a Linux-based system, your distro CD most likely already has Mesa packages (like RPM or DEB) which you can easily install.
You're application is written in IRIS GL, not OpenGL. IRIS GL was the predecessor to OpenGL and is a different thing (almost) entirely. Mesa's not the solution.
GLUT (OpenGL Utility Toolkit) is in the separate MesaGLUT-x.y.z.tar.gz file. If you don't already have GLUT installed, you should grab the MesaGLUT package and compile it with the rest of Mesa.
On Linux-based systems you'll want to follow the Linux ABI standard. Basically you'll want the following:
After installing XFree86/X.org and the DRI drivers, some of these files may be symlinks into the /usr/X11R6/ tree.
The old-style Makefile system doesn't install the Mesa libraries; it's up to you to copy them (and the headers) to the right place.
The GLUT header and library should go in the same directories.
Stand-alone Mesa (downloaded as MesaLib-x.y.z.tar.gz) doesn't have any support for hardware acceleration (with the exception of the 3DFX Voodoo driver).
What you really want is a DRI or NVIDIA (or another vendor's OpenGL) driver for your particular hardware.
You can run the glxinfo program to learn about your OpenGL library. Look for the GL_VENDOR and GL_RENDERER values. That will identify who's OpenGL library you're using and what sort of hardware it has detected.
glxinfo
If your DRI-based driver isn't working, go to the DRI website for trouble-shooting information.
Make sure the ratio of the far to near clipping planes isn't too great. Look here for details.
Mesa uses a 16-bit depth buffer by default which is smaller and faster to clear than a 32-bit buffer but not as accurate. If you need a deeper you can modify the parameters to glXChooseVisual in your code.
glXChooseVisual
Be sure you're requesting a depth buffered-visual. If you set the MESA_DEBUG environment variable it will warn you about trying to enable depth testing when you don't have a depth buffer.
Specifically, make sure glutInitDisplayMode is being called with GLUT_DEPTH or glXChooseVisual is being called with a non-zero value for GLX_DEPTH_SIZE.
glutInitDisplayMode
GLUT_DEPTH
This discussion applies to stencil buffers, accumulation buffers and alpha channels too.
Be sure you have an active/current OpenGL rendering context before calling glGetString.
If you're trying to draw a filled region by using GL_POINTS or GL_LINES and seeing holes or gaps it's because of a float-to-int rounding problem. But this is not a bug. See Appendix H of the OpenGL Programming Guide - "OpenGL Correctness Tips". Basically, applying a translation of (0.375, 0.375, 0.0) to your coordinates will fix the problem.
These can be overridden by using the --with-max-width and --with-max-height options. The two need not be equal.
Do note that Mesa uses these values to size some internal buffers, so increasing these sizes will cause Mesa to require additional memory. Furthermore, increasing these limits beyond 4096 may introduce rasterization artifacts; see the leading comments in src/mesa/swrast/s_tritemp.h.
First, join the Mesa3d-dev mailing list. That's where Mesa development is discussed.
The OpenGL Specification is the bible for OpenGL implemention work. You should read it.
Most of the Mesa development work involves implementing new OpenGL extensions, writing hardware drivers (for the DRI), and code optimization.
Unfortunately, writing a device driver isn't easy. It requires detailed understanding of OpenGL, the Mesa code, and your target hardware/operating system. 3D graphics are not simple.
The best way to get started is to use an existing driver as your starting point. For a software driver, the X11 and OSMesa drivers are good examples. For a hardware driver, the Radeon and R200 DRI drivers are good examples.
The DRI website has more information about writing hardware drivers. The process isn't well document because the Mesa driver interface changes over time, and we seldome have spare time for writing documentation. That being said, many people have managed to figure out the process.
Joining the appropriate mailing lists and asking questions (and searching the archives) is a good way to get information.
The specification for the extension indicates that there are intellectual property (IP) and/or patent issues to be dealt with.
We've been unsucessful in getting a response from S3 (or whoever owns the IP nowadays) to indicate whether or not an open source project can implement the extension (specifically the compression/decompression algorithms).
In the mean time, a 3rd party plug-in library is available.