Satish Balay <[email protected]> writes:
> Adding in the following [partial] dependencies gets the build working
> for me. But I'm not sure what the correct way to add this would be..

Was the problem reproducible?  I never encountered it.

Can we determine in the makefile whether MPIUNI is being used?  We could
generate the proper dependencies in gmakegen.py, but if we're not
checking the source to pick up module dependencies, I'd rather use the
hack in gmakefile.


> Satish
>
> ------
>
> balay@petsc^/sandbox/balay/petsc.clone((4920dd4...)) $ git diff
> diff --git a/gmakefile b/gmakefile
> index bd3ef2e..543c229 100644
> --- a/gmakefile
> +++ b/gmakefile
> @@ -134,6 +134,9 @@ $(OBJDIR)/%.o : %.cu | $$(@D)/.DIR
>         $(PETSC_COMPILE.cu) $< -o $@ # Compile first so that if there is an 
> error, it comes from a normal compile
>         @$(PETSC_GENDEPS.cu) $< -o $(@:%.o=%.d) # Generate the dependencies 
> for later
>  
> +$(OBJDIR)/src/snes/f90-mod/petscsnesmod.o:$(OBJDIR)/src/ksp/f90-mod/petsckspmod.o
> +$(OBJDIR)/src/ts/f90-mod/petsctsmod.o:$(OBJDIR)/src/snes/f90-mod/petscsnesmod.o
> +
>  $(OBJDIR)/%.o : %.F | $$(@D)/.DIR
>  ifeq ($(FC_MODULE_OUTPUT_FLAG),)
>         cd $(PETSC_ARCH)/include && $(FC) -c $(FC_FLAGS) $(FFLAGS) 
> $(FCPPFLAGS) $(FC_DEPFLAGS) ../../$< -o ../../$@

Attachment: pgpvy7HzFY9rS.pgp
Description: PGP signature

Reply via email to