Re: problem of WRAPIT

Hi,

    Thanks Mary and Rick for your kindness help. I have narrowed my data, but it's still too large as an attachment,sorry.My data is coming from WRFV2.2.

     The ex.ncl ad ex.f works well in my machine, I got the correct result.
   
    I have try one test that I call wrf_cape_2d and DCAPECALC3D2 in the same ncl script (actually,these two functions are coming from the same fortran code rip_cape.f), the wrf_cape_2d works well but the DCAPECALC3D2 failed just after the calling, the command "print*,'vvvvvvvvv'" just after constant specified in rip_cape.so(line 581) doesn't work too, though I use the same data.

     So, Maybe I should use some option when WRAPIT rip_cape.f?

     Following is the information when I doing the ncl test
_________________________________________________________________
[yangyh@gateway ncl]$ ncl cptp.ncl
 Copyright (C) 1995-2009 - All Rights Reserved
 University Corporation for Atmospheric Research
 NCAR Command Language Version 5.1.0
 The use of this software is governed by a License Agreement.
 See http://www.ncl.ucar.edu/ for more details.
(0) DX,DY = 9000 9000
(0) Working on time: 2010012505 3
(0) llz=34
(1) llz=264
(2) llz=264
(0) llz=34
(1) llz=264
(2) llz=264
(0) llz=34
(1) llz=264
(2) llz=264
(0) llz=34
(1) llz=264
(2) llz=264
(0) llt=264
(1) llt=264
(0) ll=34
(1) ll=264
(2) ll=264
(0) ll==34
(1) ll==264
(2) ll==264
(0) wrf_cape_2d begin ****************
(0) wrf_cape_2d end ****************
(0) UUUUUU UUU
Segmentation fault
________________________________________________________________

    Thanks again!!

Yuhua
  ----- Original Message -----
  From: Rick Brownrigg
  To: Yuhua Yang
  Cc: Mary Haley
  Sent: Wednesday, January 27, 2010 12:38 AM
  Subject: Re: [ncl-talk] problem of WRAPIT




  Yuhua,


  I can't see from the static sources that anything is wrong in your call to DCAPECALC3D2 from ncl. About the only thing I can suggest is to instrument the fortran code itself with write statements to dump key variables to see if what's being passed in from NCL is what you expect. You might also use write statements as waypoint markers, to narrow in on where in the code the SEGV is occurring.


  Rick


  On Jan 26, 2010, at 7:07 AM, Mary Haley wrote:


    Yuhua,


    In just looking at your program, it looks like you are calling the Fortran routine correctly. You appear to be passing in the arrays correctly, you're using the right types, and you have the right order of arguments. It also appears that you renamed all the functions in rip_cape.f, which is very important so it doesn't clash with existing functions already built into NCL.


    Maybe Rick will see something I don't!


    I'm unable to run your code because it depends on other NCL scripts and a data file I don't have.


    Can you narrow this program way down so that we can run it? The best thing to do would be to write all the variables that this function depends on to a temporary NetCDF file, and then all I'll need to run this script is this file.


    Here's how you might write the data to a NetCDF file:


       fout = addfile("mary.nc","c")
      fout->prs = prs
      fout->tmk = tmk
      fout->qvp = qvp


    etc.


    Do this for prs, tmk, qvp, zz, hgtt, and psfcd.


    Then, I'll just be able to read in this file, and call your function below:


         DCAPECALC3D2::DCAPECALC3D2(prs,tmk,qvp,zz,hgtt,psfcd,cape,cin,mii,mjj,mkzh,0,1,filin)


    THanks,


    --Mary


    On Jan 25, 2010, at 11:06 PM, Yuhua Yang wrote:


         Thanks Mary and Rick for your help very much!
         
         Now there are no problem on WRAPIT the rip_sub.f, but another problem coming. I got "Segmentation fault" when I calling the rip_sub.so, I have check my ncl script carefully but I can't find where s the error. I'm using the wrfout of WRFV2.2, and following is the information:
      _________________________________________________________________________
      [yangyh@gateway ncl]$ ncl cptp.ncl
       Copyright (C) 1995-2009 - All Rights Reserved
       University Corporation for Atmospheric Research
       NCAR Command Language Version 5.1.0
       The use of this software is governed by a License Agreement.
       See http://www.ncl.ucar.edu/ for more details.
      (0) DX,DY = 9000 9000
      (0) Working on time: 2010012505 3
      (0) llz=34
      (1) llz=264
      (2) llz=264
      (0) llz=34
      (1) llz=264
      (2) llz=264
      (0) llz=34
      (1) llz=264
      (2) llz=264
      (0) llz=34
      (1) llz=264
      (2) llz=264
      (0) llt=264
      (1) llt=264
      (0) ll=34
      (1) ll=264
      (2) ll=264
      (0) ll==34
      (1) ll==264
      (2) ll==264
      (0) UUUUUU UUU
      Segmentation fault
      ____________________________________________________________________________________________

         Could you please check the attachment cptp.ncl for me?
         
         By the way,following is the purpose of Why I using rip_sub.f :
         The "wrf_cape_2d" works well on my machine, but I'd like to get the temperature of Equilibrium level,which doesn't output by wrf_cape_2d, so I have to call the rip_sub.f in order to get extra information. Does the function wrf_cape_2d is coming from rip_cape.f?

        Thanks again!!

        Yuhua
         

        ----- Original Message -----
        From: Mary Haley
        To: Rick Brownrigg
        Cc: Yuhua Yang ; ncl-talk@ucar.edu
        Sent: Tuesday, January 26, 2010 1:10 AM
        Subject: Re: [ncl-talk] problem of WRAPIT


        Thanks for catching this Rick! I was not aware of this restriction.


        I've added this to the "special considerations" section of the WRAPIT documentation, which is a good document to read if you are having problems with WRAPIT:


           http://www.ncl.ucar.edu/Document/Tools/WRAPIT.shtml


        --Mary


        On Jan 25, 2010, at 9:43 AM, Rick Brownrigg wrote:


          Hi,


          The root of this error is that you declare the types of your two embedded functions as:


          FUNCTION TONPSADIABAT(THTE,PRS,...
                  DOUBLE PRECISION TONPSADIABAT
          ...


          FUNCTION VIRTUAL(TEMP,RATMIX)
          DOUBLE PRECISION VIRTUAL
          ...


          While this is legitimate fortran code, its causing an error in the WRAPIT parsing code. Try declaring the types of the functions with one statement, as:


          DOUBLE PRECISION FUNCTION TONPSADIABAT(THTE,PRS,...
          ...
          DOUBLE PRECISION FUNCTION VIRTUAL(TEMP,RATMIX)




          Rick




          On Jan 25, 2010, at 3:12 AM, Yuhua Yang wrote:


            Hi,

              When I WRAPIT the attachment file(rip_cape.f) which coming from Mary, I got the following problem:
            ______________________________________________________
             WRAPIT rip_cape.f
             
            WRAPIT Version: 090115
            A syntax error occurred while parsing: TONPSADIABAT
            COMPILING rip_cape.f
            LINKING
            END WRAPIT
            ________________________________________________________

             The rip_cape.so file is created, but when I try to call the routine by my NCL script , I get the warning:
            Could not find Init() in external file ./rip_cape.so, file not loaded .

            the enviroment of my machine is linux,64-bit.

            Could you give me some hint to solve this problem? Thanks very much!

            Yuhua


              <rip_cape.f>
            _______________________________________________
            ncl-talk mailing list
            List instructions, subscriber options, unsubscribe:
            http://mailman.ucar.edu/mailman/listinfo/ncl-talk



          _______________________________________________
          ncl-talk mailing list
          List instructions, subscriber options, unsubscribe:
          http://mailman.ucar.edu/mailman/listinfo/ncl-talk


      <cptp.ncl><rip_cape.f><psadilookup.dat>
      _______________________________________________
      ncl-talk mailing list
      List instructions, subscriber options, unsubscribe:
      http://mailman.ucar.edu/mailman/listinfo/ncl-talk


    _______________________________________________
    ncl-talk mailing list
    List instructions, subscriber options, unsubscribe:
    http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk