Re: [ferret_users] reading FORTRAN sequential access data - simplequestion

["William S. Kessler" <William.S.Kessler@xxxxxxxx>]

Except for the 'access=sequential' part (which I have no idea about),  
I write unformatted fortran in just the same way (an entire grid on  
one record) to read into ferret all the time, both on Sun Solaris and  
Mac OSX. Never any problem. I think Steve's comment about the need  
for the /COLUMNS= qualifier is the exactly what's needed. That is how  
Ferret knows that there is more than 1 variable/record.

I guess I can't speak to the 'type=r8' thing, either.

Billy K

On Jul 6, 2006, at 6:07 PM, tomas@ucar.edu wrote:

> Thanks for the reply Steve.
>
> I brought up the issue of the loop because in the FORTRAN code
> in the example, the writing was structure something like:
>
> do i = 1,100
>    do j = 1,100
>       do k = 1,100
>          write(10)array(i,j,k)
>       enddo
>    enddo
> enddo
>
> If I am correct, each time the write statement is executed, a new
> record is written, and along with it, an end of record "string".
>
> Now I am writing using a structure like
>
> write(10) array
>
> In this case only a single big record and single end of record  
> "string" is
> written.
>
> I am writing and reading on the same linux box, by the way.
>
> I *think* I tried the columns = appropriate number but can't be sure
> now.  I'll try again tomorrow to be sure.
>
> Bob
>
> > Hi Bob,
> >
> > FORTRAN unformatted output is not really a standard.  It is system  
> > and
> > compiler dependent how the concept of a "record" is implemented.   
> > So for
> > a start if the data are written on one system architecture and are  
> > being
> > read on another then all bets are off.
> >
> > The second thing is that the existence of the loop in your code
> > shouldn't be a factor in successfully reading the first output  
> > record.
> > Reading this first record, alone, should be your first target.   
> > See Ch2
> > Sec4.1.1.  "Records of uniform length" in the on-line users guide.
> >
> > At a glance I think the problem in your command may be that you  
> > need to
> > include /COLUMNS=`128*64*19`.  (It has been years since I have  
> > used this
> > capability.)
> >
> >   yes? file/var=hrmean/grid=gg/type=r8/format=unf/COL=`128*64*19`
> > HRMEAN.sst-2Xna-ctrl-04_001_280_emean.dat
> >
> > If this works, then move onto the time looping.  Define the time  
> > axis as
> > part of your grid, but do not change the /COLUMNS argument.  /COLUMNS
> > tells how many values are on each record as I recall.
> >
> >     - steve
> >
> > =================================
> >
> > Bob Tomas wrote:
> >
> > > Jaison,
> > >
> > > I looked at the examples and unfortunately the one provided doesn't
> > > address the
> > > case I have because the FORTRAN write is imbedded inside do loops  
> > > and
> > > single array elements are written with each "write".  In my case, I
> > > write out the whole darn array using a single write.
> > >
> > > In Ferret I tried:
> > >
> > > define axis/x=1:128:1 x128
> > > define axis/y=1:64:1 y64
> > > define axis/z=1:19:1 z19
> > > define grid/x=x128/y=y64/z=z19 gg
> > > file/var=hrmean/grid=gg/type=r8/format=unf
> > > HRMEAN.sst-2Xna-ctrl-04_001_280_emean.dat
> > >
> > > show data
> > >      currently SET data sets:
> > >     1> ./HRMEAN.sst-2Xna-ctrl-04_001_280_emean.dat  (default)
> > >  name     title                             I         J
> > > K         L
> > >  HRMEAN   HRMEAN                           1:128     1:64
> > > 1:19      ...
> > >
> > > fill hrmean[k=19]
> > > sfe: formatted io not allowed
> > > apparent state: unit 62 named
> > > ./HRMEAN.sst-2Xna-ctrl-04_001_280_emean.dat
> > > lately reading sequential formatted external IO
> > > Abort
> > >
> > > Bob
> > >
> > >
> > >
> > >
> > >
> > >
> > > On Fri, 7 Jul 2006, Jaison Kurian wrote:
> > >
> > > > Hi Bob,
> > > >         See the examples given in this mail :
> > > >
> > > > http://ferret.pmel.noaa.gov/Ferret/Mail_Archives/fu_2004/ 
> > > > msg00372.html
> > > >
> > > >
> > > > Regards,
> > > >
> > > > Jaison
> > > >
> > > >
> > > > On Thu, 6 Jul 2006, Bob Tomas wrote:
> > > >
> > > > > Hello,
> > > > >
> > > > > I tried to sort this out using the documentation but I wasn't able
> > > > > so...
> > > > >
> > > > > I created a binary file using FORTRAN using something like:
> > > > >
> > > > >        real*8 hrmean(19,64,128)
> > > > >     open(unit=10,access='sequential',form='unformatted',...)
> > > > >        write(10)hrmean
> > > > >
> > > > > If someone can provide an example, I'd be very thankful.
> > > > >
> > > > > Bob
> > --
> > º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤ø,¸¸,ø¤º°`°º¤º¤ø,¸¸,ø¤º
> >
> > Steve Hankin, NOAA/PMEL -- Steven.C.Hankin@noaa.gov
> > 7600 Sand Point Way NE, Seattle, WA 98115-6349
> > ph. (206) 526-6080, FAX (206) 526-6744