I don't think anything should be that different between 0.3.0 and 0.3.1. Could you look at the two dll's in Dependency Walker to see whether there may be some difference in which routines are exported? You also may have your PATH configured in a way that the dependency dll's are being found on one computer, but not on the other.
On Thursday, October 16, 2014 11:39:47 AM UTC-7, David Morgan wrote: > > I have written a Julia script that calls a Fortran subroutine that solves > systems of ordinary differential equations using DVode_f90, one of the > variants of Lsode, Vode, cvode etc. developed at Lawrence Livermore > Laboratory over the years. > > On my laptop from work, I have the 64 bit Windows 7 operating system and a > 64 bit version of Julia, 3.0 dated 8-20-2014. I compile the Fortran > subroutine into a dll using 64 bit gfortran that I copied from the Anaconda > installation of Python (it is a version of the 64 bit version of MinGW). > > The Julia script calls the Fortran subroutine and everything works great. > > However, when I try to run this script on my home desktop computer, I get > an error. My desktop has the 64 bit version of Windows 7 and the 64 bit > gfortran from the Anaconda installation of Python. The desktop has a 64 bit > version of Julia, version 3.1 dated 2014-09-21. The error I get is ERROR: > error compiling anonymous: could not load module CO2Inj1dInc1_gf: The > specified module could not be found. > > Is there a difference between Julia 3.0 and 3.1 that could account for > this behavior? > > By the way, I really like Julia. I was using R and the package deSolve to > solve this problem, but it was too slow and I could not figure out how to > call a Fortran subroutine in R. It can definitely be done, but Julia is > much easier. With Julia and Fortran, the problem is solved about 50 to 100 > times faster. >
