using all.vars() doesn't fit in with df$variable, syntax as you get c(df, variable).... work out how lm does it, or make asOneFormula not use all.vars() ??