> FX, have you considered rth's comments with the method you use > for computing log10(). I'll try to look over this patch tomorrow. Yes, I should have said that I have nothing to object to his comments, so the final regtesting and commit will include his (much better) fallback version of log10l(). FX