[cig-commits] r12250 - seismo/2D/SPECFEM2D/trunk

[dkomati1 at geodynamics.org]

Author: dkomati1
Date: 2008-06-17 08:47:38 -0700 (Tue, 17 Jun 2008)
New Revision: 12250

Modified:
   seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt
Log:
updated the todo list


Modified: seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt
===================================================================
--- seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt	2008-06-17 15:37:52 UTC (rev 12249)
+++ seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt	2008-06-17 15:47:38 UTC (rev 12250)
@@ -1,21 +1,4 @@
 
-- in the case of MPI runs, we should add a routine to check that the graph sent to Cuthill-McKee is not disconnected; otherwise the routine fails
-
-- in the case of MPI runs, Nicolas's code displays a lot of things coming from
-different processors to the screen; this should be fixed, and only the master
-(myrank == 0) should write to the screen
-
-- in createnum_fast.f90 we should replace the heuristic computation of the minimum
-threshold distance:
-!
-! compute the minimum typical "size" of an element in the mesh
-  xtypdist = min(xtypdist,xmaxval-xminval)
-  xtypdist = min(xtypdist,ymaxval-yminval)
-!
-with a more general algorithm based on the actual distance (i.e., using sqrt()
-instead of an estimate of the horizontal and vertical sizes only; because otherwise
-very distorted elements may lead to erroneous results)
-
 - splitting file part_unstruct.F90 in several files for clarity purpose.
 
 - improving compiling with SCOTCH (issue with header file scotchf.h which is Fortran77 legal). Having our own scotchf.h file (without the comments) is not wise.