GnuPG 1.9.92 build failures on old Linux

Werner Koch wk at
Thu Oct 12 17:35:36 CEST 2006

On Wed, 11 Oct 2006 20:18, Alain Guibert said:

> | protect-tool.c:1187: `orig_codeset' undeclared (first use this function)
> | protect-tool.c:1187: (Each undeclared identifier is reported only once
> | protect-tool.c:1187: for each function it appears in.)


--- agent/protect-tool.c        (revision 4295)
+++ agent/protect-tool.c        (working copy)
@@ -1170,7 +1170,7 @@
   char *pw;
   int err;
   const char *desc;
+#ifdef ENABLE_NLS
   char *orig_codeset = NULL;
   int error_msgno;

> | extern void   usleep __P((unsigned long __usec));
> While libpth 2.0.7 /usr/local/include/pth.h contains:
> | extern int            pth_usleep(unsigned int);
> | #define usleep        pth_usleep

That seems to be a regression in Pth.  2.0.1 requires the
definition of PTH_SYSCALL_SOFT for this define.  I'll check the lates
Pth version.

> | watchgnupg.c: In function `main':
> | watchgnupg.c:227: `socklen_t' undeclared (first use this function)

Well the usual socklen_t problem.  Will include a test to configure.

> | asschk: cmd_expect_ok: expected OK but got `ERR 50331649 General error <GpgSM>'

Without the agent this is expected.  In fact, --disable-agent is not
very useful except for package building.



More information about the Gnupg-devel mailing list