The ABINIT Users Mailing List ( CLOSED )

[Roberto Pasianot <pasianot@cnea.gov.ar>]

 Hi, this is Roberto again.

 At last I found that my compiler ( pgf90-3.3-2 ) is buggy !.
 Tried on different systems with newer and older glibc's;
 the "reshape" intrinsics is not handling SOURCE arguments
 of (rank .ge. 3), so the little code below doesn't work. 
 I should file a bug report to PGI instead ... :(

 Thanks for your time. Regards,

 Roberto

 PS: sample program that doesn't work  

        real a(3,3,3),b(27)
        do i=1,3; do j=1,3; do k=1,3
             a(i,j,k) = float(i+j+k)
        end do; end do; end do
        write(6,*)a
        b=reshape(a,(/27/))
        write(6,*)b
        end
 
---------------------------------------------------------------------------
On Thu, 20 Nov 2003, Roberto Pasianot wrote:

> 
>  Hello developers and gurus out there !,
> 
>  A few days ago I reported a failure with "make test1",
> 
>         ABINIT's version: 4.2.3 (and 4.0.4, and ...)
>         System          : Linux , Athlon processor
>         compiler        : pgi 3.3-2
>  
>         ../abinis < test1.files > test1.log
>         0: RESHAPE: not enough elements in SOURCE array
> 
>  I've found that the offending code is related to some of the
>  "reshape" sentences in ./Src_3iovars/outvars.f and did something
>  to fix them; the test is passing now. Please see below for 
>  the relevant lines. The fix consisted in adding the PAD argument 
>  set to zero.
>  
>  Now, because I'm not a F90 sage I have two questions
> 
>  1) Is that a real fix or am I breaking the code ?
>  2) Are those bugs in ABINIT or in my compiler ?
> 
>  Best regards,
> 
>  Roberto
>  
> --------------------------------------------------------------------
> 
> !symrel 
>  if(multi_nsym==0)then
>   do idtset=0,ndtset_alloc
>    intarr(1:3*3*nsym,idtset)=&
> &   reshape(dtsets(idtset)%symrel(1:3,1:3,1:nsym),(/3*3*nsym/) )
> !&  reshape(dtsets(idtset)%symrel(1:3,1:3,1:nsym),(/3*3*nsym/),(/0/) )
>   enddo
> 
> 
> !xangst
>  if(multi_natom==0)then
>   dprarr(1:3*natom,0:ndtset_alloc)=&
> &   reshape(xangst_(1:3,1:natom,0:ndtset_alloc),(/3*natom,ndtset_alloc+1/) )
> !&  
> reshape(xangst_(1:3,1:natom,0:ndtset_alloc),(/3*natom,ndtset_alloc+1/),(/0.0d0/)
>  )
> 
>  
> !xcart
>  if(multi_natom==0)then
>   dprarr(1:3*natom,0:ndtset_alloc)=&
> &   reshape(xcart_(1:3,1:natom,0:ndtset_alloc),(/3*natom,ndtset_alloc+1/) )
> !&  
> reshape(xcart_(1:3,1:natom,0:ndtset_alloc),(/3*natom,ndtset_alloc+1/),(/0.0d0/)
>  )
>   
> ---------------------------------------------------------------------------
> 
> PS:
> BTW, the info on "dtsets" contains a typo,
>             ... 
> !! INPUTS
> !!  choice= 1 if echo of preprocessed variables, 2 if echo after call driver
> !!  dtsets(0:ndtset_alloc)=<type datafiles_type>contains all input variables
>                      ****  <type datasets_type>  ****
> 
> 
> |----------------------------------------------------------------------|
> |   Roberto C. Pasianot                 Phone: 54 11 6772 7244         |
> |   Dpto. Materiales, CAC-CNEA          FAX  : 54 11 6772 7303         |
> |   Avda. Gral. Paz 1499                Email: pasianot@cnea.gov.ar    |
> |   1650 San Martin, Buenos Aires                                      |
> |   ARGENTINA                                                          |
> |----------------------------------------------------------------------|
>