Dear R-sig-hpc,
I am having trouble using sfSapply on a function that calls a C function
through the .C interface. When executing in serial I have no errors, but in
parallel I have the error that each node cannot find the function called by
.C
i.e. if I have
myfn <- function(){ .C("myfunction") }
I get the error
C symbol name "myfunction" not in load table
I use sfExportAll() and sfLibrary(mylibrary) before calling the parallel
code. Thanks so much for your help!
Sincerely,
Carl
You will also need to do:
sfClusterEval(.dynload("file.so"))
to load the shared object in each node.
Vinh