On Fri, Feb 12, 2010 at 5:35 PM, Tim Rice <[email protected]> wrote:
> Attached as cofig.log.gz and compile.log.gz

I saw a few warnings:

> "/opt/src/utils/nut-2.4.1-r2339/drivers/apcsmart.c", line 34: warning: too 
> many struct/union initializers
> "/opt/src/utils/nut-2.4.1-r2339/drivers/apcsmart.c", line 44: warning: too 
> many struct/union initializers
> "/opt/src/utils/nut-2.4.1-r2339/drivers/bcmxcp.c", line 135: warning: too 
> many struct/union initializers
> "/opt/src/utils/nut-2.4.1-r2339/drivers/bcmxcp_ser.c", line 17: warning: too 
> many struct/union initializers
> "/opt/src/utils/nut-2.4.1-r2339/drivers/belkin.c", line 39: warning: too many 
> struct/union initializers
[...]

Hopefully the preceding lines compiled properly - we use those
structures all over the place.

> "/opt/src/utils/nut-2.4.1-r2339/drivers/apcsmart.c", line 127: warning: 
> statement not reached

Is there a way to flag a statement as "not reached" with the Sun C
compiler? We have a comment to that effect in the code there.

> Some Solaris 10 build notes:
> The Solaris 10 (U8) SUNWsmcmd package is broken so
> /usr/sfw/bin/net-snmp-config is symlinked to net-snmp-config-64
> instead of net-snmp-config-32. So configure will find the SNMP bits
> but the build will fail.

I am not familiar with Solaris' 32/64 bit support. Is there some other
place in the code that we are specifying 32-bit code? Should we check
some system variable and look for net-snmp-config-32 before
net-snmp-config?

> If the symlink gets fixed there is still
> the issue of net-snmp-config spitting out "-R../lib" that the linker
> doesn't like. This post configure perl script does the trick.
> perl -pi -e "s|-R../lib |-R/usr/sfw/lib |g;" \
>    `find . -name Makefile -print`

I suspect that we would want to handle this earlier in the
configuration process, probably around when those options are
retrieved. Is there a chance that someone would see "-R../lib" with
the absolute path ending up as something other than /usr/sfw/lib? (I
am trying to find out how reasonable it would be to hardcode that
fix.)

-- 
- Charles Lepple

_______________________________________________
Nut-upsdev mailing list
[email protected]
http://lists.alioth.debian.org/mailman/listinfo/nut-upsdev

Reply via email to