Here are few questions I expect to be frequently asked about GLUT 3.7. First, here are tag-line summaries of the question subject matter.
Q1: I've tried to use the "mkmkfiles.imake" script to generate Makefiles so I can build GLUT, but it doesn't seem to work.
A1: While Imakefiles are supposted to be system independent (hence the "I"), the commands to translate Imakefiles into Makefiles varies from system to system. The X Consortium provides a command called "xmkmf", but vendors do not put this command in a consistent place. The "mkmkfiles.imake" script tries its best to generate Makefiles, but may get confused by different vendors configurations that I am not aware of.
It is also possible the imake configuration files (typically located at /usr/lib/X11/config) are buggy or from a very old version of X.
SGI users can benefit from using the "mkmkfile.sgi" script that uses SGI's parallel make, though "mkmkfiles.imake" should work too.
Q2: GLUT needs improved menus, dialog boxes, scrollbars, text entry fields, etc. to be useful to me?
A2: GLUT does not pretend to be a full-featured graphical user interface toolkit.
You _could_ write these sorts of GUI objects using GLUT and OpenGL if you needed to. The other alternative is to use Motif or whatever full featured toolkit you have.
Q3: What new things are in GLUT 3.0?
A3: See README.glut3 or read The OpenGL Utility (GLUT) Programming Interface document.
Q4: Is there a version of GLUT for Windows NT or Windows 95.
A4: Nate Robins and Layne Christensen at Evans & Sutherland has been working on a freely distributable version of GLUT for Windows 95 and NT (European mirror). His efforts are directed at porting GLUT 3.3.
Q5: Is there a version of GLUT for OS/2?
A5: Yes. I believe a version based on GLUT 2.x is distributed on an OS/2 OpenGL developer's CD-ROM.
Q6: Is there a version of GLUT for the Power Mcintosh?
A6: Was told by Template Graphics that an incomplete version of GLUT had been developed for their OpenGL product for the Power Mcintosh. I am not sure if it was ever completed or made available.
Q7: I'm hesitant about upgrading to GLUT 3.0 since I've got things working will with GLUT 2.3. Is the transition painful?
A7: I do not believe so. There are two changes worth noting that _may_ affect programs you have written.
First, you need a display callback registered before your display your windows on the screen. It did not make sense for this to not be true. In all likeihood, this should not affect your GLUT programs if they written well.
Second, you can no longer change, create, or destroy menus while pop-up menus are in use. Before, you could do this, but it meant a menu might be changed while in use. It was near impossible to describe what should happen in the case of menus being changed while in use that was likely to be portable to the way other window systems handled menus, so I made the practice illegal.
You can register a menu status callback to know when menus become used and unused to avoid changing menus while they are in use.
For more details about what has changed, see the CHANGES file.
Q8: So how do I use GLUT and Motif together?
A8: You don't. To make GLUT simple and easy-to-program, GLUT supplies its own event processing loop. This makes it nearly impossible to combine GLUT and Motif. If you want Motif, you probably want a full-featured toolkit, and you ship skip GLUT and implement your application directly in Motif.
Q9: I have a bunch of simpe OpenGL programs using the aux toolkit descibed in the OpenGL Programming Guide (the "red" book). Is there an easy way to convert them to GLUT?
A9: In the progs/redbook directory, there is a script named aux2glut.sed It will give you a good start at converting simple aux calls to their GLUT equivalents. It is a good start, but you'll still have to hand edit some things.
Here's a usage example:
sed -f aux2glut.sed < aux_prog. > glut_prog.c
Q10: I have IRIX 6.2 (or 6.1) and I'd like to write GLUT programs run in true 64-bit and/or benefit from the recent, faster MIPS processors. How do I build GLUT to support these newer application binary interfaces (ABIs)?
A10: See README.irix6
Q11: I'd like to write FORTRAN programs using GLUT and OpenGL. How do I use GLUT with FORTRAN?
A11: GLUT does have a FORTRAN language binding.
For instructions for building a binding library for Silicon Graphics workstations, see README.fortran
If you want to use GLUT and OpenGL or Mesa on with Fortran on non-SGI systems, I recommend that you check, William Mitchell's f90gl home page.
Q12: I'd like to use the sophisticated input devices that GLUT supports. What should I know about this?
A12: GLUT uses the X Input extension to talk to these devices. Because the X Input extension gives a framework for supporting input devices, but does not manadate how particular devices are supported, it is possible that each vendor supports the same input devices differently.
GLUT as implemented supports SGI's means of advertising the tablet, dial & button box, and Spaceball devices. I am not sure how other vendors support these devices. For the details of SGI's support for these devices, see README.xinput Since there is no benefit in each vendor supporting these same devices in a different an incompatible way, I encourage other vendors to implement their devices in this same manner.
Q13: Can I use GLUT and Open Inventor?
A13: Yes. See the README.inventor file. Also, some source code examples can be found at progs/inventor
Because the Open Inventor development enviornment is not supported on all systems, the Inventor example programs are not built by default, and the Makefile there only support SGI systems.
Q14: I have Sun workstation, and it is supposed to support overlays. So why does GLUT not use them?
A14: GLUT uses the SERVER_OVERLAY_VISUALS convention that advertises overlay visuals. Most major workstation vendors support this convention (DEC, HP, IBM, SGI), but Sun does not.
Q15: The stroke font used for GLUT looks familar. Where did it come from?
A15: The data for the "stroke roman" font is lifted from the X11R5 PEX sample implementation.
Q16: I read in the NOTICE file that you are writing a book on programming OpenGL for the X Window System. When will it be available?
A16: At SIGGRAPH '96 or possibly before that.
Q17: You mention an unnamed bu "very large window system software vendor" as the reason portable GLUT programs should not directly include <GL/gl.h> and <GL/glu.h> directly. What's the vendor and what are the details?
A17: Microsoft. It's version of <GL/gl.h> requires <windows.h> to be included before <GL/gl.h> can be included because of Microsoft function declaration conventions. Sigh.
Q18: I want my GLUT program to read and send information over a socket to some other program. How do I do this in in GLUT?
A18: You can not do it currently. I am considering such support for a possible GLUT 4.0. I'd like to have a portable solution.
What you'd like is a callback that would tell you when a socket is ready for reading and writing. I'm hoping to find a way to support this in an operating system independent manner. Does anyone know of a good portable interface for networked bytestream connections?
For now, you've got the source code to GLUT and you could hack it into GLUT for whatever particular interface your operating system provides.
Q19: Where's the best place to ask questions about GLUT or OpenGL? Can I just email them to you?
A19: While I may try to return email if I have time, the best place is the comp.graphics.api.opengl newsgroup. This gives a lot more people a chance to answer your question and you'll probably get an answer much faster than sending me email. Plus, I may not know the answer though someone on the "net" may know it.
Q20: My workstation doesn't have OpenGL. Where can I get a free copy to use with GLUT?
A20: OpenGL is licensed by Silicon Graphics and is not available as "free" or "public domain" software, but workstation vendors typically bundle OpenGL software with their workstation. However, there is a package called Mesa written by Brian Paul at the University of Wisconsin that implements the OpenGL API. (To be branded as "OpenGL", an implementation must be licensed and pass the Architectural Review Board's conformance suite, so Mesa is not an official "OpenGL" implementation.) Mesa does work with GLUT.
Q21: I hear GLUT 3.0 has overlay support. Where is an example?
A21: Look at progs/examples/zoomdino.c for an example of using overlays for rubber-banding and display of a help message, both in the overlays. Also, test/over_test.c exercises all of the overlay routines.
Q22: I get BadMatch X protocol errors when I run GLUT programs. What gives?
A22: There is a bug in the Solaris 2.4 and 2.5 implementation of XmuLookupStandardColormap (fixed in Solaris 2.6). When you compile GLUT on Solaris 2.4 or 2.5, please apply the following patch and compile with -DSOLARIS_2_4_BUG to workaround the problem. To do this, edit the glut/lib/glut/Makefile and add -DSOLARIS_2_4_BUG to the CFLAGS macro. See the comment in the patch below. This code is already in GLUT 3.1 and later.
*** glut_win.c Wed Apr 24 14:06:08 1996
--- glut_win.c.bad Wed Apr 24 14:03:58 1996
***************
*** 398,414 ****
case TrueColor:
case DirectColor:
*colormap = NULL; /* NULL if RGBA */
- #ifndef SOLARIS_2_4_BUG
- /* Solaris 2.4 has a bug in its XmuLookupStandardColormap
- implementation. Please compile your Solaris 2.4 version
- of GLUT with -DSOLARIS_2_4_BUG to work around this bug.
- The symptom of the bug is that programs will get a
- BadMatch error from X_CreateWindow when creating a GLUT
- window because Solaris 2.4 creates a corrupted
- RGB_DEFAULT_MAP property. Note that this workaround
- prevents Colormap sharing between applications, perhaps
- leading unnecessary colormap installations or colormap
- flashing. */
status = XmuLookupStandardColormap(__glutDisplay,
vi->screen, vi->visualid, vi->depth, XA_RGB_DEFAULT_MAP,
/* replace */ False, /* retain */ True);
--- 398,403 ----
***************
*** 423,429 ****
return;
}
}
- #endif
/* If no standard colormap but TrueColor, just make a
private one. */
/* XXX Should do a better job of internal sharing for
--- 412,417 ----
Q23: What is new in GLUT 3.1?
A23: GLUT 3.1 is largely a maintence release. There are some new programs, a few minor GLUT library bug fixes, but mostly GLUT 3.1 is to make sure GLUT builds cleanly on various platforms like SunOS, HP/UX, Solaris, and Linux. See the CHANGES file included in the distribution for more details.
Q24: How do I make Linux shared libraries for GLUT?
A24: Peter F. Martone (pmarton@mailbox.bgsu.edu) has written some instructions for making a Linux shared library for GLUT. You can grab the instructions for doing so from http://pizza.bgsu.edu/cgi-bin/cgiwrap/~pmarton/makeMainIndex
Q25: New in GLUT 3.2.
A25: Like GLUT 3.1, GLUT 3.2 is a maintence release. Along with bug fixes to the core GLUT library, many new GLUT example programs have been added. The portability of the examples has been improved so that most should build using Windows 95 and NT. Also, GLUT API man pages are now included. See the CHANGES file included in the distribution for more details.
Q26: GLUT API man pages.
A26: Please see the README.man file for details. The easiest way for SGI users to get the man pages is to install the "glut_dev.man.glut" subsystem included with the pre-compiled SGI GLUT images.
Q27: Fast window repair for Mesa.
A27: The GLX specification states that the state of a window's back color buffer after a glXSwapBuffers is undefined. However, the freeware Mesa implementation of the OpenGL API always leaves the back buffer with its previous contents (ie, it simply "copies" the back buffer contents to the front buffer).
Because Mesa lacks hardware acceleration and is often slow to redraw a window, this presents the opportunity to speed redrawing a window damaged by window system interactions by simply calling glXSwapBuffers again.
If you set the MESA_SWAP_HACK enviornment variable, GLUT 3.2 will try to repair double buffered windows not otherwise needing a redisplay because of glutPostRedisplay by calling glXSwapBuffers when Mesa is the OpenGL implementation being used and the last display callback called glutSwapBuffers.
In general, this means if you see MESA_SWAP_HACK when using Mesa, double buffered GLUT programs will redraw very quickly after being damaged but still operate well if they've been correctly written to use glutPostRedisplay to trigger application required redraws.
I encourage all Mesa users to set the MESA_SWAP_HACK environment variable.
Q28: Advanced GLUT example .rgb image file.
A28: Yes, the image files these examples use are large and were seperated out from the main GLUT source code distribution. Get the glut_data.tar.gz file from where you got your GLUT distribution. Untar these data files over your glut distribution so the "data" directory is at the same level as "progs". Then do a "make links" in the progs/advanced directory to make symbolic links.
See the progs/advanced/README file for more details.
Q29: Why doesn't GLUT programs compiled on IRIX 6.4 or 6.3 work earlier releases?
A29: First, SGI never guarantees that an executable built on a later IRIX release will work on an earlier release. Sometimes it works; more often than not it does not. GLUT takes advantage of a new X optimization in IRIX 6.3 called "fast atoms". This optimization lets X clients determine common atom values without an X server round-trip. This helps X performance.
If you compile the GLUT library on an IRIX 6.3 or IRIX 6.4 machine, the library will support fast atoms. This will mean that if you run executables linked against the "fast atom enabled" version of the GLUT library, you'll get a run-time link error saying something like:
17062:glut_example: rld: Fatal Error: attemped access to unresolvable symbol in projtex: _XSGIFastInternAtom
Do not be alarmed. If you want, you can recompile the GLUT library with the -DNO_FAST_ATOMS and get a version of the library that doesn't have the support so that GLUT executables built with a library compiled without "fast atoms" can work on earlier IRIX releases. Note that even if you do compile with -DNO_FAST_ATOMS, there is still no guarantee that an IRIX executable compiled on a newer release will actually work on an older release (but at least you'll have a chance!).
Note that the precompiled images lack "fast atoms" support so they will work fine with IRIX releases before IRIX 6.3 and 6.4.
Q30: Can I get a version of GLUT for the Power Macintosh?
A30: Probably pretty soon. Conix Graphics is working on a port of GLUT 3.2 as of late January 1997. Try checking the Conix Graphics web site http://www.conix3d.com for current info.
Q31: What is new in GLUT 3.4?
A31: GLUT 3.4 is an incremental release. An Ada binding for SGI machines is included along with an Ada example. Many new sample programs. Several such as dinoshade.c demonstrate real-time rendering techniques relevant for games. Examples using Sam Leffler's libtiff library for loading, drawing and writing TIFF image files. GLUT version of the facial animation "geoview" decibed in the Parke and Water's book "Computer Facial Animation". New API interfaces to be made part of the GLUT 4 API update (not yet fully finalized though). glutInitDisplayMode for example. Improved portability and a few bug fixes.
Q32: I installed SGI's Cosmo3D beta and GLUT, and I'm having problems compiling GLUT programs.
A32: Unfortunately, SGI's Cosmo3D beta images install a DSO for GLUT (libglut.so) that does not fully implement the GLUT API and lacks some of the newer GLUT 3.4 entrypoints as well. The problem is that a DSO takes preferenc over an archive when you compile with an option like "-lglut". While the Cosmo3D beta installs a libglut.so, my GLUT distribution and images only build and install an archive. There are a couple of solutions:
su cd /usr/lib mv libglut.so libglut.so.cosmo cc -32 -o libglut.so -shared -all libglut.a cd /usr/lib32 mv libglut.so libglut.so.cosmo cc -n32 -o libglut.so -shared -all libglut.a
The new DSO generated from the GLUT 3.4 DSO should be compatible with the old Cosmo version. This will mean that all the GLUT programs you build will need the libglut.so on the machine they run on.
Q33: What is new in GLUT 3.5?
A33: The most significant change with GLUT 3.5 is unifying the X Window System and Win32 versions of GLUT into a single source code distribution. Henk Kok contributed several cool new demos (rollercoaster, chess, opengl_logo). All the demos build cleanly under Win32. Lots of bug fixes. Interesting new OpenGL rendering techniques are demonstrated in a number of new examples: movelight, dinoshade, halomagic, rendereps, movelight, shadowfun, torus_test, underwater, texfont, reflectdino.
Q34: How do I use the precompiled Win32 GLUT DLLs with Borland compilers?
A34: The "implib" command should let you generate a GLUT.LIB that works with Borland compilers from the precompiled GLUT.DLL Here is an example:
C:\>implib C:\GLUT\LIB\GLUT.LIB C:\WINDOWS\SYSTEM\GLUT.DLL
After this, then link C:\GLUT\LIB\GLUT.LIB to your project
Suggested by Carter <carter@extremezone.com>.
Q35: Are there any C++ wrappers for GLUT?
A35: Yes, George Stetten (stetten@acpub.duke.edu) of Duke University has made available the GlutMaster C++ wrapper classes. See:
http://www.duke.edu/~stetten/GlutMaster/GlutMaster.html http://www.duke.edu/~stetten/GlutMaster/README.txt
Q36: How do you avoid the Console window appearing when you compiler a Win32 GLUT application with Microsoft compilers?
A36: Try using the following Microsoft Visual C compiler flags:
/SUBSYSTEM:WINDOWS /ENTRY:mainCRTStartup
These are linker options... if main or wmain are defined, MSVC build a CONSOLE app by default; hence the need for /SUBSYSTEM:WINDOWS. if /SUBSYSTEM:WINDOWS is defined, MSVC expects WinMain or wWinMain to be defined; hence the need to /ENTRY:mainCRTStartup (eg the entry point is the usual C main).
stdout/stderr are [apparently] not "attached"; output via printf is simply "eaten" unless redirected at the command-line or by a parent program.
Information thanks to Jean-David Marrow (jd@riverbed.com).
Q37: What is new in GLUT 3.6?
A37: GLUT 3.6 adds/improves the following:
See the "CHANGES" file that accompanies GLUT 3.6 for a fuller list of changes.
Q38: On my IRIX 6.3 SGI O2 workstation, why do I get errors about "glXChannelRectSyncSGIX" being unresolved building certain GLUT examples?
A38: The original IRIX 6.3 release for the O2 workstation accidently advertised support for the dynamic video resize extension supported on SGI's high-end InfiniteReality graphics system. This confuses GLUT into providing its dynamic video resize sub-API.
This problem is fixed by patch 1979 (and its successor patches). Because patch 1979 (and its successor patches) also help O2's OpenGL rendering performance, I strongly recommend requesting the latest O2 OpenGL patch from SGI customer support.
Once the patch is installed, your build errors will be resolved.
Q39: Using GLUT with Microsoft OpenGL 1.1 and compiling GLUT with Borland compilers causes GLUT applications to generates floating point exceptions. What can be done?
A39: Under certain conditions (e.g. while rendering solid surfaces with lighting enabled) MS libraries cause some illegal operations like floating point overflow or division by zero. The default behaviour of Microsoft compilers is to mask (ignore) floating point exceptions, while Borland compilers do not. A function of Borland run-time library allows to mask exceptions. Modify glut_init.c by adding the following lines to the function __glutOpenWin32Connection:
#ifdef __BORLANDC__ #include_control87(MCW_EM,MCW_EM); #endif
With this modification, compiling the GLUT library with your Borland compilers and using GLUT with Microsoft OpenGL should work fine.
GLUT 3.7 will have this change already included in the GLUT library source code distribution.
This advice comes from Pier Giorgio Esposito (mc2172@mclink.it).
Q40: Using GLUT with SGI OpenGL for Windows and compiling with Borland compilers results in linking problems. What can be done?
A40: Some care must be taken when linking GLUT.DLL or programs that use it with Borland compilers. The import library IMPORT32.LIB already contains the functions exported by the Microsoft OpenGL libraries, thus SGI OpenGL import libraries must be listed _before_ import32 in the Borland tlink command line.
This advice comes from Pier Giorgio Esposito (mc2172@mclink.it).
Q41:What is GameGLUT?
A41: GameGLUT is a set of API extension to GLUT to be released in GLUT 3.7. These extensions provide keyboard release callbacks, disabling of keyboard auto repeat, joystick callbacks, and full screen resolution setting.
Questions, send mail to mjk@nvidia.com