User: rhaas Date: 2012/06/21 06:36 PM
Modified: /trunk/ interface.ccl
Log: GRHydro: change prolongation type of real valued atmosphere mask to none
this fixes the aborts by Carpet for vertex centered simulations in which it complained about the wrong order of prolongation being used.
Me culpa, since I suggested using "copy" rather than "none".
File Changes:
Directory: /trunk/ ==================
File [modified]: interface.ccl Delta lines: +1 -1 =================================================================== --- trunk/interface.ccl 2012-06-21 16:06:48 UTC (rev 365) +++ trunk/interface.ccl 2012-06-21 23:36:47 UTC (rev 366) @@ -427,7 +427,7 @@ # This real mask is set during UpdateAtmosphereMask and sync'ed afterwards (including possible interpatch interpolation) # After syn'ing and before any con2prim fun, we set the integer mask above based on the real-valued mask. # This ensures that any routine using the int mask is still correctly working. -real GRHydro_atmosphere_mask_real type = GF Timelevels = 1 tags='Prolongation="copy" checkpoint="no"' +real GRHydro_atmosphere_mask_real type = GF Timelevels = 1 tags='Prolongation="none" checkpoint="no"' { atmosphere_mask_real } "Flags to say whether a point needs to be reset to the atmosphere. This is sync'ed (and possibly interpolated)!"