[scponly] New scponly Snapshot Release

Sat Nov 20 11:18:42 EST 2010

On Sat, Nov 20, 2010 at 3:44 AM, Frank Fegert <fra.nospam.nk at gmx.de> wrote:
> Hello,
>
> thank you very much for the new snapshot!
>
> On Fri, Nov 19, 2010 at 10:56:14PM -0800, Kaleb Pederson wrote:
>> I've just released a snapshot that I could use some help testing.  If you are
>> in a position that enables you to test, I'd appreciate any feedback you may
>> have, good or bad.
>
> I have some issues which are most likely only AIX-/xlC-specific:
>  $ cc -qversion
>  IBM XL C/C++ for AIX, V10.1
>  Version: 10.01.0000.0005
>
>  $ gmake
>  cc -qmaxmem=16384 -DSYSV -D_AIX -D_AIX32 -D_AIX41 -D_AIX43 -D_AIX51 -D_AIX52 -D_AIX53 -D_AIX61 -D_ALL_SOURCE -DFUNCPROTO=15 -I/opt/freeware/include -I. -I. -Wall -pedantic -DHAVE_CONFIG_H -DDEBUGFILE='"/opt/freeware/etc/scponly/debuglevel"' -o scponly.o -c scponly.c
>  cc: 1501-210 (S) command option Wall contains an incorrect subargument
>  gmake: *** [scponly.o] Error 40
>
> The "-Wall" option is gcc specific, xlC on the other hand uses:
>  -w     Suppresses informational, language-level and warning messages.
>         Specifying this option is equivalent to specifying -qflag=e:e
>         in C or -qflag=s:s in C++.

Frank is right, I accidentally left "-Wall" and "-pedantic" in
Makefile.in which usually fail any time the compiler isn't gcc. I know
better and just missed removing some common flags I add for testing on
my machine. For those of you who build scponly using a non-gcc
compiler, just remove those two options from Makefile.in and it should
work.

> The next problem after this is:
>  $ gmake
>  cc -qmaxmem=16384 -DSYSV -D_AIX -D_AIX32 -D_AIX41 -D_AIX43 -D_AIX51 -D_AIX52 -D_AIX53 -D_AIX61 -D_ALL_SOURCE -DFUNCPROTO=15 -I/opt/freeware/include -I. -I. -DHAVE_CONFIG_H -DDEBUGFILE='"/opt/freeware/etc/scponly/debuglevel"' -o scponly.o -c scponly.c
>  "scponly.h", line 46.1: 1506-485 (S) Parameter declaration list is incompatible with declarator for inline.
>  "scponly.h", line 47.1: 1506-276 (S) Syntax error: possible missing '{'?
>  gmake: *** [scponly.o] Error 1
>
> This can be fixed with:
>  $ diff -u scponly.h.orig scponly.h
>  --- scponly.h.orig  2010-11-20 07:31:33.000000000 +0100
>  +++ scponly.h   2010-11-20 12:23:55.000000000 +0100
>  @@ -43,8 +43,8 @@
>    /*
>     * function prototypes
>     */
>  - inline char *strbeg (char *, char *);
>  - inline char *strend (char *, char *);
>  + char *strbeg (char *, char *);
>  + char *strend (char *, char *);
>    int valid_chars(char *);
>    int get_uservar(void);
>    void show_usage(void);
>
> but i'm not sure if this breakes something which needs the strings to
> be defined inline.

No, it doesn't.  I actually remember running into this when I did have
an AIX machine available.  I've removed them and the changes are in
CVS.

> Then some warnings, just FYI:
>  $ gmake clean
>  $ gmake
>  cc -qmaxmem=16384 -DSYSV -D_AIX -D_AIX32 -D_AIX41 -D_AIX43 -D_AIX51 -D_AIX52 -D_AIX53 -D_AIX61 -D_ALL_SOURCE -DFUNCPROTO=15 -I/opt/freeware/include -I. -I. -DHAVE_CONFIG_H -DDEBUGFILE='"/opt/freeware/etc/scponly/debuglevel"' -o scponly.o -c scponly.c
>  cc -qmaxmem=16384 -DSYSV -D_AIX -D_AIX32 -D_AIX41 -D_AIX43 -D_AIX51 -D_AIX52 -D_AIX53 -D_AIX61 -D_ALL_SOURCE -DFUNCPROTO=15 -I/opt/freeware/include -I. -I. -DHAVE_CONFIG_H -DDEBUGFILE='"/opt/freeware/etc/scponly/debuglevel"' -o helper.o -c helper.c
>  cc -qmaxmem=16384 -DSYSV -D_AIX -D_AIX32 -D_AIX41 -D_AIX43 -D_AIX51 -D_AIX52 -D_AIX53 -D_AIX61 -D_ALL_SOURCE -DFUNCPROTO=15 -I/opt/freeware/include -I. -I. -DHAVE_CONFIG_H -DDEBUGFILE='"/opt/freeware/etc/scponly/debuglevel"' -o netbsd_getopt_long.o -c netbsd_getopt_long.c
>  "netbsd_getopt_long.c", line 247.48: 1506-280 (W) Function argument assignment between types "const char*" and "int" is not allowed.
>  "netbsd_getopt_long.c", line 259.39: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.
>  "netbsd_getopt_long.c", line 259.51: 1506-280 (W) Function argument assignment between types "const char*" and "int" is not allowed.
>  "netbsd_getopt_long.c", line 282.47: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.
>  "netbsd_getopt_long.c", line 282.59: 1506-280 (W) Function argument assignment between types "const char*" and "int" is not allowed.
>  "netbsd_getopt_long.c", line 373.39: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.
>  "netbsd_getopt_long.c", line 373.46: 1506-280 (W) Function argument assignment between types "const char*" and "int" is not allowed.
>  "netbsd_getopt_long.c", line 382.47: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.
>  "netbsd_getopt_long.c", line 382.54: 1506-280 (W) Function argument assignment between types "const char*" and "int" is not allowed.
>  "netbsd_getopt_long.c", line 414.47: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.
>  "netbsd_getopt_long.c", line 428.39: 1506-280 (W) Function argument assignment between types "int" and "const char*" is not allowed.

These were for warnings generated by invalid getopt parameters. I
removed all of these last night after creating the snapshot.

> The install process fails at changing ownership to root:root if one
> builds as a non-root user, which is IMHO pretty common for (RPM|deb)
> packaging:
>  ...
>  + gmake DESTDIR=/home/$USER/rpmbuild/rpms/scponly-4.8-instroot install
>  echo "0" > debuglevel
>  ./install-sh -c -d /home/$USER/rpmbuild/rpms/scponly-4.8-instroot/opt/freeware/bin
>  ./install-sh -c -d /home/$USER/rpmbuild/rpms/scponly-4.8-instroot/opt/freeware/man/man8
>  ./install-sh -c -d /home/$USER/rpmbuild/rpms/scponly-4.8-instroot/opt/freeware/etc/scponly
>  ./install-sh -c -o 0 -g 0 scponly /home/$USER/rpmbuild/rpms/scponly-4.8-instroot/opt/freeware/bin/scponly
>  chown: /home/$USER/rpmbuild/rpms/scponly-4.8-instroot/opt/freeware/bin/#inst.9764974#: Operation not permitted.
>  gmake: *** [install] Error 1
>  Bad exit status from /home/$USER/rpmbuild/rpms/rpm-tmp.3464 (%install)

I'll have to think about a good way to fix this as I'm not sure how
this is typically handled. If anybody knows, I'd be glad to
incorporate changes.

> Now i'm off to do some actual functionality testing ;-)

Great!

> HTH & best regards,

It helps a lot, thank you Frank.

--Kaleb