[534] in Kerberos-V5-bugs
Re: building under solaris v2.3
daemon@ATHENA.MIT.EDU ("Mark W. Eichin")
Tue Jun 28 22:59:36 1994
Date: Tue, 28 Jun 94 22:59:11 EDT
From: eichin@MIT.EDU ("Mark W. Eichin")
To: "Paul Pomes" <P-Pomes@uiuc.edu>
Cc: krb5-bugs@MIT.EDU
In-Reply-To: "[533] in Kerberos-V5-bugs"
In specific, platform-specific conditionals for isode go in
isode/h/config.h; you'll note things like #ifdef __FreeBSD__ in
there... since we'll be eliminating isode altogether in a patchlevel
"Real Soon Now" it isn't worth doing it right.
In general, there shouldn't *be* SYS5 or POSIX tests, only
feature-by-feature tests. If you find something that the autoconf test
"discovers" incorrectly or doesn't test at all: if it is local to one
directory, you probably want to add it to configure.in in that dir; if
it is more common, top level aclocal.m4 might be better. (Better
documentation will be forthcoming as it stabilizes; we will be
simplifying this part somewhat...)
As for solaris 2.3: This release builds under solaris 2.3 as of
patchlevel 1... and shouldn't need any tweaks. However, that is using
gcc, which defines __svr4__, which triggers this:
#ifdef __svr4__
/* SYS5 is for termio instead of sgttyb */
#define SYS5
/* SVR4 turns off strdup */
#define SVR4
#endif
Perhaps the compiler you're using doesn't define __svr4__, in which
case you should just drop it into CCOPTS...
_Mark_
