I have a Sun Blade 100 system (it is old, but works great) that I've installed Solaris 10 on. I have also installed Gnu make 3.80 and CMake 2.6.4. gcc came pre-installed and is at version 3.4.3. Whenever I follow the instructions to compile Allegro on this system, I get the error messages indicated in the attached text file. (CompileError.txt)

The cmake .. command works great.
The make command does not and is what is output in the attached text file.

The compilation seems to work up to the 16% mark, then errors out. Refer to the attached text file.

I have not modified any Allegro source code. The warnings and errors you see in the text file are what I've been getting "out of the box" when trying to compile Allegro.

Any comments or suggestions are GREATLY appreciated.

As per my first post I'm using the gcc compiler that installed with Solaris. It is gcc version 3.4.3. However, I have also tried using the SunStudio 12 compiler. Both fail when following the Allegro build instructions.

Looking at the error output (text file attached in first post) it appears that Allegro is redefining some OpenGL variables that are already defined in header files that Solaris installs.

From attached file:
..../Allegro/allegro-5.0.10/include/allegro5/opengl/GLext/glx_ext_defs.h:84:1:
warning: "GLX_SAMPLES_ARB" redefined
In file included from /usr/include/GL/glx.h:33

In the above, /usr/include/GL/glx.h came with Solaris 10. Is it normal for Allegro to redefine OpenGL variables that are already defined in OpenGL header files in the OS?

Okay, I've upgraded the system to use gcc/g++ 4.8.2. The results? EXACTLY the same.

I'll try installing a newer version of cmake onto the system. I see that version 2.8.12 is now available for download. If this doesn't work then I'm not quite sure what to do.

What versions of Gnu make, gcc/g++ and cmake are others using? Thanks!!

I just tried this with gcc 4.8.2 on Linux

#include <stdio.h>

typedef unsigned char _Bool;

int main(void)
{
  _Bool mybit = 1;
  printf("mybit is %u\n",mybit);
  
  return 0;
}

and it bombed out with

pepsi@fractal:~ 07:59 PM $ gcc -s -O2 -Wall t.c -o t
t.c:3:23: error: two or more data types in declaration specifiers
 typedef unsigned char _Bool;
                       ^
t.c:3:1: warning: useless type name in empty declaration [enabled by default]
 typedef unsigned char _Bool;
 ^
pepsi@fractal:~ 07:59 PM $

Removing the prepended underscore fixed it. Maybe manually patching it will fix it, and possibly the other errors are some type of cascade?

[EDIT]

Renaming it to _Booll also caused it to compile without error. Adding the option -std=c89 didn't affect the _Bool causing an error.
idontevenknowwhatimdoing

[EDIT2]

Eliminating the typedef entirely allowed it to compile even with -std=c89, as if it's still using _Bool as a keyword. Compiler bug? I ran into another gcc 4.8.2 bug a few weeks ago concerning structs.

It would be awesome if you contributed a patch once you're done. Then we might be able to add Unix to the list of supported platforms. Which would be cool.

Have you figured out why you're getting OpenGL errors?

Also, threads lock after a week of inactivity, so if you need us to bump the thread we will.

It sounds really weird to me that GCC on Solaris would mess up something like that. Are you sure the standard C headers are up to date too? The required header is stdbool.h, the #define is __bool_true_false_are_defined.