Determining platform type/OS in gcc via #defines

1,095 registered users | 1 active user | 2 LpH | Statistics
Login | Create New User

Welcome to IWETHEY!

Post #73,500 by admin 1/8/03 7:48:54 PM Reply	Determining platform type/OS in gcc via #defines Is there a .h to include, or some sort of built-in #define that I can use to determine chip type and OS in gcc? Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,505 by Another Scott 1/8/03 8:26:13 PM Reply	Maybe Mozilla source would help... Or maybe not. There's a web page with browsable code, and a CPU routine [link\|http://lxr.mozilla.org/seamonkey/source/gc/boehm/os_dep.c\|here]. It's pretty messy and is likely far more elaborate than you need... Going up the tree to .../boehm/ gives the other files in the build. Good luck. Cheers, Scott.
Post #73,538 by admin 1/8/03 9:34:57 PM Reply	Yes, this is the sort of thing I was looking for: [link\|http://lxr.mozilla.org/seamonkey/source/gc/boehm/gcconfig.h#27\|http://lxr.mozilla.o...ehm/gcconfig.h#27] Thanks. I'll try it out tomorrow at work. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,509 by deSitter 1/8/03 8:35:32 PM Reply	Is this on Linux? Make system do a uname -a. -drl
Post #73,519 by boxley 1/8/03 8:54:28 PM Reply	what he said will work for cash and other incentives [link\|http://home.tampabay.rr.com/boxley/resume/Resume.html\|skill set] You think that you can trust the government to look after your rights? ask an Indian
Post #73,537 by admin 1/8/03 9:33:28 PM Reply	I wasn't clear enough. I need this to happen at compile time. Currently, the build system generates -D entries based on things like uname. I would like to determine OS/platform (endian stuff, really, but I can get that from endian.h on unix platforms) based on built-in #defines produced by the compiler or present in system .h files. I don't want to have to detect the OS in the build system. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,545 by deSitter 1/8/03 9:59:56 PM Reply	Gotta be a #pragma right? You basically tell the compiler "do this if this is the environment..." -drl
Post #73,548 by admin 1/8/03 10:03:15 PM Reply	#define Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,568 by boxley 1/8/03 11:17:19 PM Reply	/usr/include/machine/cpu.h (like that?) will work for cash and other incentives [link\|http://home.tampabay.rr.com/boxley/resume/Resume.html\|skill set] You think that you can trust the government to look after your rights? ask an Indian
Post #73,570 by admin 1/8/03 11:21:24 PM Reply	Doesn't exist on this box. Or were you just giving an example of what you thought I was looking for? Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,571 by boxley 1/8/03 11:27:50 PM Reply	osx has 3 entries /usr/include/machine/cpu.h * @APPLE_LICENSE_HEADER_END@ / #ifndef _BSD_MACHINE_CPU_H_ #define _BSD_MACHINE_CPU_H_ #if defined (__ppc__) #include "ppc/cpu.h" #elif defined (__i386__) #include "i386/cpu.h" #else #error architecture not supported #endif #endif / _BSD_MACHINE_CPU_H_ / as you can see it is a stub so you might be able to write one out and add it to the build. /usr/include/i386/cpu.h @APPLE_LICENSE_HEADER_END@ / / * HISTORY * / #ifndef _BSD_I386_CPU_H_ #define _BSD_I386_CPU_H_ #include <sys/appleapiopts.h> #ifdef __APPLE_API_OBSOLETE #define cpu_number() (0) #endif / __APPLE_API_OBSOLETE / #endif / _BSD_I386_CPU_H_ */ hope anything helps, you might want to write one thanx, bill will work for cash and other incentives [link\|http://home.tampabay.rr.com/boxley/resume/Resume.html\|skill set] You think that you can trust the government to look after your rights? ask an Indian
Post #73,573 by admin 1/8/03 11:39:26 PM Reply	Heh. No /usr/include/i386 either. But yes, that's the sort of thing I'm looking for. Thanks. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,533 by ChrisR 1/8/03 9:24:43 PM Reply	I don't really have an answer handy... ...but the question is whether you want that information at compile time or at run time?
Post #73,539 by admin 1/8/03 9:35:48 PM Reply	Compile time, without passing -D switches. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,542 by ChrisR 1/8/03 9:41:52 PM Reply	If memory serves me correct... ...those -D switches are really just macro constants in disguise. So you should be able to use the constants in conditional compilation tests.
Post #73,544 by admin 1/8/03 9:53:26 PM Reply	Right. But I don't want to pass -D switches. :-) That's what I'm doing now; ie. on Linux I add -D_LINUX to the gcc command line, and look for #ifdef _LINUX in the code. I'm looking for a predefined macro constant so I don't have to pass the -D switches any more. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,588 by static 1/9/03 4:14:56 AM Reply	Could autoconf help? I've seen lot of programs do build-type checks in autoconf. Wade. Microsoft are clearly boiling the frogs.
Post #73,608 by admin 1/9/03 8:19:36 AM Reply	We have our own build system; autoconf not a part of it Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,689 by neelk 1/9/03 1:24:15 PM Reply	Autoconf could still be useful The way it works is that it runs a bunch of tests to figure out what the environment is. Even if you have your own system, you could download autoconf and steal the tricks they are using.
Post #73,691 by admin 1/9/03 1:34:45 PM Reply	Autoconf does so at runtime. ie. through uname, etc. Which isn't particularly useful during compile time. :-) Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,712 by Arkadiy 1/9/03 3:42:14 PM Reply	Are we talking about "configure" script? As used by scores of Open Source apps? It works _before_ compile time. -- We have only 2 things to worry about: That things will never get back to normal, and that they already have.
Post #73,714 by admin 1/9/03 3:46:40 PM Reply	You misunderstand me. The techniques that configure/autoconf use are runtime techniques, and hence useless to my goal of doing the configuration at compile time. :-) Basically, I ended up using stuff like: \n#if defined(MSC_VER)\n// windows stuff\n#elif defined(linux) \|\| defined(__linux__)\n// linux stuff\n#elif defined(sun) \|\| defined(__sun__)\n// solaris stuff\n#else\n#error Unknown or unsupported OS\n#endif\n Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,717 by ChrisR 1/9/03 4:04:20 PM Reply	One man's compile time... ...is another man's run time. How about having the make file run a configuration procedure that generates a header file with the constants in it. Something with pipes or grep or awk or.... :-)
Post #73,724 by admin 1/9/03 4:38:25 PM Reply	I already do that; I don't want to. :-) Or at least, something similar. I want the entire detection to happen in the source code, not as the result of something that runs before the compile. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #73,728 by ChrisR 1/9/03 4:46:21 PM Reply	So what you're really wanting is... ...to know if the compiler itself is autodetecting the environment and whether that detection is available in the conditional compilation?
Post #73,735 by admin 1/9/03 5:01:51 PM Reply	s/wanting/using Here's what I came up with, which seems to work fine: [link\|http://z.iwethey.org/forums/render/content/show?contentid=73714\|http://z.iwethey.org...w?contentid=73714] Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."

Determining platform type/OS in gcc via #defines - (admin) - (25) - Jan. 8, 2003, 07:48:54 PM EST

Maybe Mozilla source would help... - (Another Scott) - (1) - Jan. 8, 2003, 08:26:13 PM EST

Yes, this is the sort of thing I was looking for: - (admin) - Jan. 8, 2003, 09:34:57 PM EST

Is this on Linux? - (deSitter) - (8) - Jan. 8, 2003, 08:35:32 PM EST

what he said -NT - (boxley) - Jan. 8, 2003, 08:54:28 PM EST

I wasn't clear enough. - (admin) - (6) - Jan. 8, 2003, 09:33:28 PM EST

Gotta be a #pragma right? - (deSitter) - (1) - Jan. 8, 2003, 09:59:56 PM EST

#define -NT - (admin) - Jan. 8, 2003, 10:03:15 PM EST

/usr/include/machine/cpu.h (like that?) -NT - (boxley) - (3) - Jan. 8, 2003, 11:17:19 PM EST

Doesn't exist on this box. - (admin) - (2) - Jan. 8, 2003, 11:21:24 PM EST

osx has 3 entries - (boxley) - (1) - Jan. 8, 2003, 11:27:50 PM EST

Heh. No /usr/include/i386 either. - (admin) - Jan. 8, 2003, 11:39:26 PM EST

I don't really have an answer handy... - (ChrisR) - (13) - Jan. 8, 2003, 09:24:43 PM EST

Compile time, without passing -D switches. -NT - (admin) - (12) - Jan. 8, 2003, 09:35:48 PM EST

If memory serves me correct... - (ChrisR) - (11) - Jan. 8, 2003, 09:41:52 PM EST

Right. But I don't want to pass -D switches. :-) - (admin) - (10) - Jan. 8, 2003, 09:53:26 PM EST

Could autoconf help? - (static) - (9) - Jan. 9, 2003, 04:14:56 AM EST

We have our own build system; autoconf not a part of it -NT - (admin) - (8) - Jan. 9, 2003, 08:19:36 AM EST

Autoconf could still be useful - (neelk) - (7) - Jan. 9, 2003, 01:24:15 PM EST

Autoconf does so at runtime. - (admin) - (6) - Jan. 9, 2003, 01:34:45 PM EST

Are we talking about "configure" script? - (Arkadiy) - (5) - Jan. 9, 2003, 03:42:14 PM EST

You misunderstand me. - (admin) - (4) - Jan. 9, 2003, 03:46:40 PM EST

One man's compile time... - (ChrisR) - (3) - Jan. 9, 2003, 04:04:20 PM EST

I already do that; I don't want to. :-) - (admin) - (2) - Jan. 9, 2003, 04:38:25 PM EST

So what you're really wanting is... - (ChrisR) - (1) - Jan. 9, 2003, 04:46:21 PM EST

s/wanting/using - (admin) - Jan. 9, 2003, 05:01:51 PM EST

iwethey.org

Would you like some Chocolate Foam with that Mercedes, sir?
105 ms