Two bugs were found in version 2.1k in the implementation of the Voigt function
when calculating line absorption and the calculation of line broadening.
The Voigt function bug affected primarily lines with small damping parameters,
and resulted in non-fatal numerical errors in the xstar output absorption spectrum (INFs).
When xstar was called as part of xstar2xspec this resulted in fatal errors
because the cfitsio routines which read the xstar output could not interpret the INFs.
The line broadening bug resulted in too large absorption line depths when turbulent
broadening was important. Neither of these bugs affected the temperature,
ionization balance or emission spectrum. The bugs have been repaired in version 2.1kn3.
Version 2.1kn3 also has an added feature, which is the addition of ion-by-ion
heating and cooling rates as extensions to the output file xout_abund1.fits.
Also added is the capability to set the value of niter to a negative number, which
allows the solution of charge conservation without solving thermal equilibrium. As before,
if niter=0 then neither charge transfer nor thermal equlibrium is calculated.