#9 Compile broken

[Vitaly Lipatov]

I get an error
scanner.l: In function `yylex':
scanner.l:156: error: `UINT32_MAX' undeclared
(first use in this function)
during compile last cvs and fix it with patch:

--- scanner.l.o 2005-07-17 15:58:33 +0400
+++ scanner.l 2005-07-17 15:57:53 +0400
@@ -12,6 +12,7 @@
#include <stdio.h>
#include <string.h>
#include <math.h>
+#include <stdint.h>
#include <errno.h>
#include <ctype.h> /* for isdigit() */
#include "calculator.h"

[Kyle Wheeler]

Logged In: YES
user_id=17596

Err? That's not what the latest CVS looks like... But, your point is taken,
and I'm now testing the compile on several OS's: Debian Linux, MacOS X,
OpenBSD, and Solaris, and I'll add any extra portability code to it to get it
to compile on those systems.

What system are you running?

[Vitaly Lipatov]

Logged In: YES
user_id=930161

It is ALT Linux.
$ uname -a
Linux server.office.etersoft.ru 2.6.12-std26-up-alt1 #1
Mon Jul 4 17:37:57 MSD 2005 i686 unknown
unknown GNU/Linux

$ rpm -q glibc
glibc-2.3.5-alt5

What header provides UINT32_MAX you guess?

[Kyle Wheeler]

Logged In: YES
user_id=17596

Well, it depends on the system. On OpenBSD, for example, UINT32_MAX
isn't provided at all (neither is stdint.h). In the current CVS (which is, for
the moment, identical to the 2.0 source), I have several tests. If stdint.h
exists, it is included. Solaris does not have a stdint.h, but instead has
many of those integer types in inttypes.h - which I test for and include if it
exists. And then, finally, I put in there that if UINT32_MAX was not defined
anywhere, to define it myself. Hopefully that addresses the issue. Can you
compile 2.0 on your computer without modification?

[Vitaly Lipatov]

Logged In: YES
user_id=930161

Thank you, it is fine now.