Message-ID: <Pine.A41.4.58.0405030719240.127142@homer12.u.washington.edu>
Date: 2004-05-03T16:24:45Z
From: Thomas Lumley
Subject: plot functions, formula interfaces and NAs
In-Reply-To: <Pine.LNX.4.44.0405031412001.26494-100000@gannet.stats>
On Mon, 3 May 2004, Prof Brian Ripley wrote:
> I believe no function should be using getOption("na.action") without a
> means to override it, and thus the formula methods of
>
> boxplot, mosaicplot and pairs
>
> should get an na.action argument. Further, since in all cases the
> non-formula methods do handle missing values, I think the default should
> be na.pass. That is a change in behaviour, but the existing behaviour is
> undocumented.
>
As I know Brian realises, but other people may not, it is not sufficient
just to add na.action=na.pass as a default argument in a function that
uses the usual match.call/eval.parent method to handle formulas. The
default will be overridden by the default in model.frame.
-thomas