[Met_help] MET installation problem

John Halley Gotway johnhg at ucar.edu
Mon Feb 22 07:50:47 MST 2010


Hmmm, yes.  I do see the same type of behavior as well.  I found that xlC really was just pointing to xlc (see my directory listing below).

However, when I tried rebuilding MET using xlc, I got a bunch of compilation errors with undefined references!  For more details, I've attached three files to this message:
(1) The Makefile I used that successfully compiles MET using xlC.
(2) The log file when successfully building MET using xlC.
(3) The log file showing the error messages I got when trying to build with xlc.

So I'm not sure what exactly the difference is between xlc and xlC, but there does actually appear to be a difference.

Here's a long listing of the IBM compilers on the NCAR/bluefire machine:

be1005en% ls -la /usr/bin/xl*
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC128 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC128_r -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC128_r4 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC128_r7 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC_r -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC_r4 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlC_r7 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc128 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc128_r -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc128_r4 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc128_r7 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc_r -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc_r4 -> /usr/vac/bin/xlc
lrwxrwxrwx    1 root     system           16 Jul 03 2008  /usr/bin/xlc_r7 -> /usr/vac/bin/xlc
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf2003
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf2003_r
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf90
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf90_r
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf90_r7
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf95
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf95_r
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf95_r7
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf_r
-rwxr-xr-x   16 bin      bin          191639 Sep 30 13:07 /usr/bin/xlf_r7

Litta A J wrote:
> Dear sir,
> 
> Sorry for the delay.. I didn't come to the department on last two days...
> 
> I am trying to install GNU c/c++ to the IBM-AIX5.3. I have told to IBM
> people about xlC.. They didnt send any reply...
> 
> I have searched in our 16 processor IBM system for xlC and found one link of
> xlC..  but the link connect to the c compiler. I have checked that in
> another 4 and 64 processor IBM machine, then i found the same link for xlC.
> Can you please check in your IBM machine where it is linking?
> 
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128_r4@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128core@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128core_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC128core_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC_r4@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlC_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlCcore@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlCcore_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlCcore_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128_r4@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128core@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128core_r@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++128core_r7@ ->
> /usr/vac/bin/xlc*
> lrwxrwxrwx   1 bin      bin              16 Oct 14 2008  xlc++_r@ ->
> /usr/vac/bin/xlc*
> with best regards,
> Litta
> On Fri, Feb 19, 2010 at 9:13 PM, John Halley Gotway <johnhg at ucar.edu> wrote:
> 
>> Litta,
>>
>> Sorry, but I don't have much more to offer.  I have only compiled MET on an
>> IBM using the xlC (C++ compiler) and the xlf (Fortran) compiler.  I'd guess
>> that since you have xlc (C compiler) available,
>> you should be able to get the xlC C++ compiler as well, but I really have
>> no idea about that.
>>
>> If you by any chance also have access to a LINUX machine, you could always
>> try building MET there as well.
>>
>> Thanks,
>> John
>>
>> Litta A J wrote:
>>> Dear sir,
>>>
>>> Thank you very much for your kinfd information.. But I didn't see any C++
>>> compilers in IBM system. I have asked to my system administrator about
>> the
>>> C++ compiler. He told that, C++ compiler is not there and he don't know
>> how
>>> to install. I am attaching the list of all C compilers in my system..I'll
>>> contact to IBM people,with the help of my guide, from where we have
>>> purchased the system...
>>>
>>> thunder>pwd
>>> /usr/vac/bin
>>> thunder>ls
>>> CreateExportList*  c99_128_r*         cc_r*              xlc128*
>>> c89*               c99_128_r4*        cc_r4*             xlc128_r*
>>> c89_128*           c99_128_r7*        cc_r7*             xlc128_r4*
>>> c89_128_r*         c99_r*             cleanpdf*          xlc128_r7*
>>> c89_128_r4*        c99_r4*            fixpkg_vacndi*     xlc_r*
>>> c89_128_r7*        c99_r7*            gxlc*              xlc_r4*
>>> c89_r*             cc*                mergepdf*          xlc_r7*
>>> c89_r4*            cc128*             resetpdf*          xlchelp*
>>> c89_r7*            cc128_r*           showpdf*
>>> c99*               cc128_r4*          vacndi*
>>> c99_128*           cc128_r7*          xlc*
>>> thunder>
>>>
>>> with best regards,
>>> Litta
>>>
>>> On Thu, Feb 18, 2010 at 9:42 PM, John Halley Gotway <johnhg at ucar.edu>
>> wrote:
>>>> Litta,
>>>>
>>>> I noticed something right away in the information you sent...  It looks
>>>> like you're using the wrong compilers.
>>>>
>>>> xlc is a C compiler, and xlC is a C++ compiler.  It's pretty annoying
>> that
>>>> the only difference is the lower-case vs upper-case letter 'c', but
>> that's
>>>> the difference between them.  And you should be
>>>> using the C++ compiler here.
>>>>
>>>> Try setting:
>>>> CXX = /usr/vac/bin/xlC
>>>> FC  = /usr/vac/bin/xlf
>>>>
>>>> And then be sure to do a "/path/to/your/make clean" before trying to
>>>> rebuild.  That should help a lot.
>>>>
>>>> Thanks,
>>>> John
>>>>
>>>> Litta A J wrote:
>>>>  > Dear sir,
>>>>>  Thank you very much for your kind information...  You are right. It is
>>>>> making libraries after typing as you specified.. It is successfully
>>>> creating
>>>>> vx_afm Library, vx_analysis_util Library, vx_cal Library files. But it
>> is
>>>>> not creating vx_color library. I am attaching make.log file and the
>>>>> specified lines of stdlib.h Please go through that and give your
>> valuable
>>>>> suggestion.... My CXX path is  as follows...
>>>>>
>>>>> # Path to the C++ Compiler
>>>>> # C++ compiler flags
>>>>> # Any additional required libraries
>>>>> CXX          = /usr/vac/bin/xlc
>>>>> CXX_FLAGS    = # -g -qfullpath
>>>>> CXX_LIBS     =
>>>>>
>>>>> thunder>/user2/litta/MET_T/make-3.81/make
>>>>> *** Making the Model Evaluation Tools Project ***
>>>>>
>>>>> *** Making vx_afm Library ***
>>>>> /usr/vac/bin/cc afm.cc   -c -I..
>>>>> /usr/vac/bin/cc afm_keywords.cc   -c -I..
>>>>> /usr/vac/bin/cc afm_line.cc   -c -I..
>>>>> /usr/vac/bin/cc afm_token.cc   -c -I..
>>>>> /usr/vac/bin/cc afmkeyword_to_string.cc   -c -I..
>>>>> /usr/vac/bin/cc afmtokentype_to_string.cc   -c -I..
>>>>> ar -rs libvx_afm.a afm.o afm_keywords.o afm_line.o afm_token.o
>>>>> afmkeyword_to_string.o afmtokentype_to_string.o
>>>>> ar: Creating an archive file libvx_afm.a.
>>>>> ranlib libvx_afm.a
>>>>> cp libvx_afm.a ../.
>>>>> *** Making vx_analysis_util Library ***
>>>>>
>>>>> *** Making vx_cal Library ***
>>>>>
>>>>> *** Making vx_color Library ***
>>>>> /usr/vac/bin/cc vx_color.cc   -c -I..
>>>>> "/usr/vacpp/include/cstdlib", line 72.15: 1540-0403 (S) "abs(long)" is
>>>>> already defined.
>>>>> "/usr/vacpp/include/stdlib.h", line 146.32: 1540-0425 (I) "abs" is
>>>> defined
>>>>> on line 146 of "/usr/vacpp/include/stdlib.h".
>>>>> "/usr/vacpp/include/stdlib.h", line 38.32: 1540-0424 (I) "abs" is
>>>> declared
>>>>> on line 38 of "/usr/vacpp/include/stdlib.h".
>>>>> make[3]: *** [vx_color.o] Error 1
>>>>> make[2]: *** [vx_color_all] Error 2
>>>>> make[1]: *** [libs] Error 2
>>>>> make: *** [all] Error 2
>>>>> Looking forward to your reply..............
>>>>>
>>>>> with best regards,
>>>>> Litta
>>>>>
>>>>>
>>>>> On Wed, Feb 17, 2010 at 8:52 PM, John Halley Gotway <johnhg at ucar.edu>
>>>> wrote:
>>>>>> Litta,
>>>>>>
>>>>>> I went to an IBM machine here and tried building MET using the *WRONG*
>>>>>> version of make, not the GNU make.  When I did that, I saw the exact
>>>> same
>>>>>> error messages you're seeing.  Based on the command
>>>>>> line you sent me, it looks like you're just typing "make" on the
>> command
>>>>>> line.  Doing that tells the shell to search your path for "make", and
>>>> I'm
>>>>>> guessing it's still finding the wrong version of it -
>>>>>> and not the one that you installed in your home directory.  Instead,
>>>> you'll
>>>>>> need to type something like this:
>>>>>>
>>>>>> thunder> cd METv2.0
>>>>>> thunder> /home/litta/path/to/make
>>>>>>
>>>>>> Where that path specifies the location of the GNU make executable you
>>>> just
>>>>>> built.
>>>>>>
>>>>>> Let me know if you continue to experience problems.
>>>>>>
>>>>>> Thanks,
>>>>>> John
>>>>>>
>>>>>>
>>>>>>
>>>
>>>
> 
> 
> 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile
Url: http://mailman.ucar.edu/pipermail/met_help/attachments/20100222/cfa4a27f/attachment-0003.pl 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: make_met_xlc.log
Url: http://mailman.ucar.edu/pipermail/met_help/attachments/20100222/cfa4a27f/attachment-0004.pl 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: make_met_xlC.log
Url: http://mailman.ucar.edu/pipermail/met_help/attachments/20100222/cfa4a27f/attachment-0005.pl 


More information about the Met_help mailing list