On Fri, Feb 28, 2003 at 03:17:00PM -0600, Eric wrote:
> I noticed this today when trying to compile 0.9.9 -- the host OS
> is OpenBSD 3.2-stable. Thanks in advance for any insight.
>
> In file included from packet-ncp2222.c:23376:
> packet-ncp2222.inc: In function `print_nds_values':
> packet-ncp2222.inc:1674: warning: comparison between signed and
> unsigned
> packet-ncp2222.inc: In function `process_multivalues':
> packet-ncp2222.inc:3547: warning: comparison between signed and
> unsigned
> packet-ndps.c: In function `ndps_req_hash_cleanup':
> packet-ndps.c:1774: warning: unused variable `request_value'
> packet-rmi.c: In function `dissect_ser':
> packet-rmi.c:225: warning: unused parameter `pinfo'
> packet-rmi.c: At top level:
> packet-rmi.c:60: warning: `ett_rmi_protocol' defined but not used
> packet-rmi.c:68: warning: `ett_ser_magic' defined but not used
> packet-rmi.c:69: warning: `ett_ser_version' defined but not used
Those are warnings of problems not related to the compile problem you're
having.
> ld: -ldocsis: no match
It appears that libtool and/or the configure script might believe that
run-time dynamic linking is impossible on OpenBSD 3.2. I'd be skeptical
of any such conclusion on their part, although *if* OpenBSD 3.2 uses
a.out rather than ELF on whatever type of machine you're using, there
*might* be a problem with "dlopen()" requiring leading underscores on
symbols, but GLib's module support, which is what we use for dynamic
loading, should work around that.
Could you get rid of the 0.9.9 tree you have, re-download, save the
output of the configure script to a file, and send us the output of the
configure script?
Also, does a message like
libtool: link: warning: `AC_LIBTOOL_DLOPEN' not used. Assuming
no dlopen support.
appear anywhere in the output of the make? (It appears in the output of
a build here on Solaris 2.8/SPARC, with automake 1.5, autoconf 2.57, and
libtool 1.3.4.;