id	summary	reporter	owner	description	type	status	priority	milestone	component	version	resolution	keywords	cc
266	Corrections to ropp_abel_tool etc	Ian Culverwell	Ian Culverwell	"ROPP6.0 beta testing uncovered some errors in the '''ropp_pp_{abel,invert}_{exp,lin}.f90''' subroutines. Most of these appeared because the beta tester's compiler failed on SQRT(<0), whereas all ours produced !NaNs which were processed OK. 

These have been fixed, and the beta tester has run the code successfully on his machine. 

As well as removing the SQRT(<0) bugs by generating ropp_MDFV instead of !NaNs, other improvements have been made:

   * Cleaning out of the '''test2.nc''' file, which contains the results of the _LIN routines, instead of continually appending to it; 

   * Consistent use of linear extrapolation to estimate the starting bangle and dlogn/dx in the _LIN routines;

   * Correction to the upper bounds of some integrals (this led to many of the SQRT(<0)s);

   * Correction to the expressions for the bangle and refrac (use of variables at lower bound: failure to do this can lead to discontinuities in bangle curve); 

   * General tidying up where needed.
"	defect	closed	normal	6.0	ropp_pp	5.0	fixed	abel, invert