Just for kicks I have attempted to compile NASA’s GISTEMP source code. I downloaded a fresh copy of the code, and I collected the archival versions that I have downloaded previously since the code was first posted in 2007. I use Windows :). I am writing this post on a system running Windows 7 Beta x64 Ultimate. The code documentation lists the OS requirements as Unix-like. OK. The Unix-like environments that I have are the Windows ‘native’ Subsystem for Unix Applications that is the current implementation of the old NT POSIX subsystem. Great, I’m using the Beta, so using the SUA isn’t supported yet :(. I have the Cygwin environment that emulates *nix. I have several Ubuntu Linux distros running under Microsoft Virtual PC 2007 and Sun’s Virtual Box.
First Attempt: Cygwin environment. Use export command to set an environment variable named FC to use g95. No joy. Syntax errors galore. Shell errors too. What Fortran compiler did they use? Documentation’s unclear. GISS compiler command list shows f90, ifort, and xlf90 -qfixed. What Fortran standard are they using? FORTRAN IV/66, FORTRAN 77, Fortran 90, Fortran 95, Fortran 20xx? Hm, given the list, probably F77 or F90. Looking at the *.f files it looks like F77. Nice, I guess NASA’s still using code first written in the mid 80’s that has had a few student’s summer job and/or minty fresh postdoc’s ministrations tacked on over the years.
Second attempt: Cygwin environment. Set the FC environment variable to g77. Oh boy, more of the same syntax and shell errors noted above.
Third and fourth attempts: Cygwin environment. Set the FC variable back to g95. Use the second oldest and most recent archival copies. Same result as before. Compile fails but the syntax errors are gone. Shell errors only.
Fifth and sixth attempts: Cygwin environment and FC set to g77. No joy. Results are the same as the previous two attempts.
Seventh attempt: Cygwin environment and FC set to g95. Use the fresh copy. Nice, the syntax errors are back.
Eighth attempt: Cygwin environment and FC set to g77 using the fresh copy. Same result as #7.
Sigh. Is this one of the bases used to justify my being robbed in taxes to pay for, and threated with jail or death if I dare question the soundness of the premises used to scare me into accepting tyranny and slavery; or called a denier [implied moral equivalence to a Holocaust Denier like little Hugo Chavez or Mahmoud Ahmadinejad]?
Next set of four attempts: Cygwin environment, FC set to gfortran. Wonderful, same syntax errors (9 and 12) and shell errors (9 to 12).
Okay, time to look again at the suggested compiler list. Absoft’s f90 is $$$ with a time limited free trial period whether we are going to use straight Windows or Linux. I decided not to try to install the free trial into Cygwin. Next is Intel’s ifort. Hey, it’s free for personal use in a Linux environment. I’ll try this one. Oops, can’t install it in Cygwin. Off to Ubuntu for you to be checked out later. Finally, there’s IBM’s XL Fortran. Gee, $$$ with a time limited free trial. No, I don’t think I’ll try to download you and attempt the Cygwin install. Looks like the Cygwin install for these will only be by building from source.
Back to the drawing board. I’ll have to try manual compilation of each step’s sub-step executables. Who thinks that this will be a good use of time outside of academia? And I’ll have to try to fix the shell scripts to get rid of the shell diagnostic messages. I’d rather spend the time learning R than tinkering with bash. Off to Ubuntuland, got to get Jaunty Jackalope up and running.
Update 5/5/09: I have Ubuntu 9.04 up. I installed gfortran, g95, and the personal use version of Intel’s Fortran Compiler 11.0.083. No joy in getting GISTEMP to compile. More of same script errors in bash or Fortran syntax errors. I wonder what the magic specs, shell options, and compiler options were used to get it to run.
–“It never ceases to amaze me that people, especially leftist politicians, think that you can wave a magic wand and repeal the laws of physics, chemistry, and economics.”
]]>
. Yes, you are looking at snow. It is more likely to snow here in March than December, ‘Global Warming’ notwithstanding.