Test WRAPIT(Version: 051007) on intel Mac 10.4 with gfortran and g95

From: guangshan Chen <gchen9_at_nyahnyahspammersnyahnyah>
Date: Sun, 09 Dec 2007 00:33:26 -0600

Dear NCL world,

First, I want to give thanks to Matt Fearon who sent email to the
list and talked about the WRAPIT and gfortran. I got some scripts
from his attachments and changed a little bit on the base of his
scripts.

Using his scripts, I tested WRAPIT (Version: 051007) on My intel Mac
book (Mac OS 10.4) and got some interesting things that I want to
report to the list.

1. Testing WRAPIT using gfortran
As Mary's suggestion, WRAPIT works well after adding -l gfortran
options.

Here is the result:
[monsoon:~/wrapit_test] guangsha% test_90.sh

WRAPIT Version: 051007
COMPILING tester.f90
LINKING
END WRAPIT

  Copyright (C) 1995-2007 - All Rights Reserved
  University Corporation for Atmospheric Research
  NCAR Command Language Version 5.0.0
  The use of this software is governed by a License Agreement.
  See http://www.ncl.ucar.edu/ for more details.

Variable: ng
Type: integer
Total Size: 4 bytes
             1 values
Number of Dimensions: 1
Dimensions and sizes: [1]
Coordinates:
           10

2. Testing WRAPIT using g95
WRPIT using g95 works fine. But It comes out some weird things like
"indirect jmp without `*'".

Here is the result:
[monsoon:~/wrapit_test] guangsha% test_90.sh

WRAPIT Version: 051007
/sw/bin/g95
COMPILING tester.f90
/var/tmp//ccvVla10.s:46:indirect jmp without `*'
/var/tmp//ccvVla10.s:65:indirect jmp without `*'
/var/tmp//ccvVla10.s:84:indirect jmp without `*'
/var/tmp//ccvVla10.s:99:indirect jmp without `*'
LINKING
END WRAPIT

  Copyright (C) 1995-2007 - All Rights Reserved
  University Corporation for Atmospheric Research
  NCAR Command Language Version 5.0.0
  The use of this software is governed by a License Agreement.
  See http://www.ncl.ucar.edu/ for more details.

Variable: ng
Type: integer
Total Size: 4 bytes
             1 values
Number of Dimensions: 1
Dimensions and sizes: [1]
Coordinates:
  10

After some googling, I found this web site:
http://coding.derkeiler.com/Archive/Fortran/comp.lang.fortran/2006-08/
msg01227.html
(Thanks to this guy)

It is said that the problem was caused by having a newer version of
the apple assembler (I don't know what it is for), an 'as -v'
returns. It could be solved by adding '-fno-pic' to the command. But
this method did not work for him. Fortunately, It works for me.

Here is the testing:
3.1. compiling f90 file using g95 without -fno-pic
[monsoon:~/wrapit_test] guangsha% g95 -o test_90 test_90.f90
/var/tmp//cceIkhz5.s:71:indirect jmp without `*'
/var/tmp//cceIkhz5.s:90:indirect jmp without `*'
/var/tmp//cceIkhz5.s:109:indirect jmp without `*'
/var/tmp//cceIkhz5.s:124:indirect jmp without `*'
/var/tmp//cceIkhz5.s:143:indirect jmp without `*'

3.2 compiling f90 file using g95 with the option
[monsoon:~/wrapit_test] guangsha% g95 -fno-pic -o test_90 test_90.f90
[monsoon:~/wrapit_test] guangsha%

You see that the weird informations in case 3.1 are gone in case 3.2.

So I think the weird information in case 2 would be gone if we could
add "-fno-pic" to the g95 command during building the NCL 5.0.
The WRAPIT works with g95, but I don't like that kind informations
come out during compiling. Could anybody rebuild the binary file of
NCL 5.0 on ESG with the option I suggest and tell me how do you build
NCL 5.0 on intel Mac OS 10.4? Because I tried building it yesterday
and failed.

Thanks,

Guangshan

_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sat Dec 08 2007 - 23:33:26 MST

This archive was generated by hypermail 2.2.0 : Tue Dec 11 2007 - 14:19:21 MST