Hi Graham, Just a brief comment on this issue you raised: > Fortran offers no way to tell a routine how to send arguments, only how to > receive them. So when I say, for example > nt_open = nc_open(path,modev,ncid) > I cannot explain to nc_open that all three arguments are pointers. It expects > path and ncid as pointers but modev as a value. The NetCDF code has elaborate > machinery for sorting out this kind of thing, but so far at least I have > failed to understand it. > > With a 64-bit C compiler I could solve this problem by compiling 64-bit > objects equivalent to the 32-bit objects produced by gcc and understood by my > older 32-bit program. The Fortran 2003 standard (F03) provides a standardized mechanism for interoperating with C functions and types. For practical purposes, all current Fortran compilers implement this part of the F03 standard, which provides how to send arguments fro Fortran to C as well as how to receive values from C functions. A good description of these capabilities is in this document, in Chapter 5, "Interoperability with C": ftp://ftp.nag.co.uk/sc22wg5/N1551-N1600/N1579.pdf or in the examples in this guide "Standardized Mixed Language Programming for Fortran and C": http://www.roguewave.com/DesktopModules/Bring2mind/DMX/Download.aspx?entryid=751&command=core_download&PortalId=0&TabId=607 The netCDF Fortran version 4.4-beta1 or the newer development version available from svn ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-fortran-4.4-beta1.tar.gz http://svn.unidata.ucar.edu/repos/netcdf-fortran/trunk/ implement the F03 standard for calling C from Fortran, so would be a better basis for a Windows port than the current netcdf-4.2 release, in my opinion. We'll be working from the current svn trunk for our next netCDF-Fortran release, which we hope to make available in the next few months, if we don't encounter any unanticipated difficulties, and we intend that it will work on Windows. --Russ > -----Original Message----- > From: Unidata netCDF Support [mailto:address@hidden > Sent: 19 February 2013 19:02 > To: address@hidden > Cc: address@hidden > Subject: [netCDF #HFH-711245]: Linking from Windows Fortran to NetCDF 4.2 > shared library > > Good afternoon Graham, > > > I am trying to rebuild, in a 64-bit Windows 7 command-line environment using > > Intel Visual Fortran compiler XE 12.1, an executable that used to work fine > > with NetCDF 3.5 in a 32-bit Windows environment. I formerly used gcc to > > compile the C code and Intel Fortran for my own Fortran code. I have tried > > and failed to build the MinGW-64 C compiler, and it seems that I must invest > > in a commercial 64-bit C compiler. One way to avoid this, that ought to work > > but doesn't, is to rely on your shared library from > > NetCDF-4.2.dev-NC4-64-shared.exe. > > > > I'll try my best to help you out. A couple of things to start; first, > beginning with > netCDF 4.2, the netCDF-C and netCDF-Fortran libraries have been separated > from each other. The netCDF-Fortran source code is available for download > from: > > http://www.unidata.ucar.edu/downloads/netcdf/netcdf-fortran/index.jsp > > The netCDF-Fortran libraries depend on the netCDF-C libraries, so those will > need to be installed first. > > The pre-built libraries you reference, netCDF-4.2.dev-NC4-64-shared, built > using a developer snapshot; these are netCDF-C libraries. The are built > specifically for use with Visual Studio and/or in a development environment > using the Microsoft development libraries. Work to provide corresponding > netCDF-Fortran libraries is ongoing, but they are not yet available. > > The netCDF-C and corresponding netCDF-Fortran libraries can be used on > Windows when using the MSYS/MinGW or Cygwin > development environments. > > You mention the difficulties you've had building the 64-bit MinGW C compiler; > it is certainly true that it is not very intuitive. According to my notes, I > was able > to build MinGW-64 using the code available at: > > http://sourceforge.net/projects/mingw-w64/. > > Once downloaded, I have a note to follow the directions in 'mingw-w64-v2.0.7/ > mingw-w64-doc/howto-build'. Note that I built MinGW-64 inside of the MSYS > environment, installed from here: http://www.mingw.org/wiki/MSYS > > If you were to go this route, I can confirm that the netCDF-C and > netCDF-Fortran > libraries appear to compile with MinGW. Likewise, if you were able to use > cygwin, the netCDF-C and netCDF-Fortran libraries are available as packages > which can be installed from the cygwin package manager. > > I'm sorry I can't point you to an MFC/Windows-developer-library compatible > netCDF-fortran library, but hopefully this information will help. > > Have a great evening, > > -Ward > > p.s., I notice that this support request is entered in our system twice; I > will > close the earlier one, but leave this one open if there are any questions I > can help with. > > > > > > With thanks for your attention. > > > > > > > > Graham Cogley. > > > > > > > > J. Graham Cogley, Ph.D., Professor of Geography, > > > > Department of Geography, Trent University, > > > > Peterborough, Ontario, CANADA K9J 7B8. > > > > > > > > Tel +1 705-748-1011-x7686 > > > > Email address@hidden > > > > Web http://www.trentu.ca/geography/glaciology > > > > > > > > > > > > Ticket Details > =================== > Ticket ID: HFH-711245 > Department: Support netCDF > Priority: Normal > Status: Open > > Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: HFH-711245 Department: Support netCDF Priority: Normal Status: Closed
NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.