[ET Trac] [Einstein Toolkit] #350: Autogenerating cctk_Loop.h

[Einstein Toolkit]

#350: Autogenerating cctk_Loop.h
--------------------------+-------------------------------------------------
  Reporter:  eschnett     |       Owner:        
      Type:  enhancement  |      Status:  review
  Priority:  minor        |   Milestone:        
 Component:  Cactus       |     Version:        
Resolution:               |    Keywords:        
--------------------------+-------------------------------------------------

Comment (by hinder):

 2. I think it would be good to have OpenMP support in this, as then we can
 say that Cactus supports OpenMP out of the box, without adding the
 complexity of LoopControl, which is really designed to solve a much harder
 problem.
 3. You mean variable names for each?  The macro then declares variables
 named in this way for these things.  What is the reason for letting the
 user choose these variable names, rather than providing a standard, such
 as CCTK_NORMAL?  Maybe your way is better.  I think it should be
 consistent for all the macros.  The current implementation does not do
 everything that I said above; you cannot determine how far you are from
 the boundary.  What do you think about this?
 4. The usual case is to loop over "internal boundaries", so I think these
 should be called just "boundaries".  Isn't it quite unusual to want to
 loop over ghost points and symmetry points?  That version should have the
 less common name.  Maybe we can find a way around the backwards
 compatibility issue.  This code has never been released (it was committed
 after the release in November), and I suspect might not have been used.
 "internal boundaries" sounds like it loops over all ghost zones, as those
 are internal to the domain.

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/350#comment:12>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit