[ET Trac] [Einstein Toolkit] #1800: suspicious use of reflevel and refinementlevel in CarpetIOHDF5's AddAttributes function

[Einstein Toolkit]

#1800: suspicious use of reflevel and refinementlevel in CarpetIOHDF5's
AddAttributes function
--------------------+-------------------------------------------------------
 Reporter:  rhaas   |       Owner:  eschnett           
     Type:  defect  |      Status:  new                
 Priority:  minor   |   Milestone:                     
Component:  Carpet  |     Version:  development version
 Keywords:          |  
--------------------+-------------------------------------------------------
 AddAttributes in CarpetIOHDF5 takes an explicit argument
 {{{refinementlevel}}} which is used to populate the "level" attribute of
 HDF5 datasets. However it also directly uses Carpet's global variable
 {{{reflevel}} to access some internal Carpet datastructures.

 Right now this seems to only be able to cause an issue for grid arrays
 (not grid functions) that have a coordinate system attached to them (since
 the code is behind an if statement that depends on the result of
 {{{Coord_GroupSystem (cctkGH, groupname)}}}). For such an array the
 {{{pos}}} ivect would be computed use the baseextents appropriate for a
 grid function (which is wrong) on the current refinement level (which is
 also wrong since grid arrays currently exits on reflevel 0 only). Output
 for grid arrays and scalar happens in global mode, which during the time
 output happens coincides with the finest refinement level.

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1800>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit