On Mar 12, 2012, at 1:57 PM, Matthew Knepley wrote:

> On Mon, Mar 12, 2012 at 1:41 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
> 
> On Mar 12, 2012, at 1:36 PM, Jed Brown wrote:
> 
> > On Mon, Mar 12, 2012 at 13:30, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> > A couple maintenance issues that I think should be addressed.
> >
> > 1. Namespace EVERYTHING in conf/variables and 
> > $PETSC_ARCH/conf/petscvariables. It should be safe to include these files 
> > in someone else's makefile without conflicting with their own stuff.
> >
> > 2. Rename all C++ files to *.cxx or whatever so that it's possible to build 
> > C++ plugins/implementations, with everything else built using a C compiler.
> >
> > 3. Move all private headers from include/private/ to include/petsc/ 
> > (convention adopted by hundreds of packages) or include/petsc-private/ 
> > (like Tk).
> 
>  I prefer petsc-private because if one installs in the xxxx/petsc/  you get 
> this stupid looking  xxxx/petsc/include/petsc.   Better 
> xxxx/petsc/include/petsc-private
> 
> I agree that petsc/include/petsc is stupid. I see no reason to change the 
> perfectly logical include/private just to match what
> some idiot programmers did.

   I agree with Jed here. The problem now is if you install PETSc you get 
something like

   /usr/include/private/vecimpl.h   etc etc.   Someone just looking at that 
directory doesn't know it is associated with petsc while 
/usr/include/petsc-private/vecimpl.h makes that clear.

   Barry

> 
>    Matt
>  
> 
>   Barry
> 
> 
> 
> 
> 
> -- 
> What most experimenters take for granted before they begin their experiments 
> is infinitely more interesting than any results to which their experiments 
> lead.
> -- Norbert Wiener


Reply via email to