[Rd] unwarranted warning from hist.default (PR#9356)

[deepayan.sarkar at gmail.com]

On 10 Nov 2006 23:39:14 +0100, Peter Dalgaard <p.dalgaard at biostat.ku.dk> wrote:
> deepayan.sarkar at gmail.com writes:
>
> > > x = rnorm(100)
> > > b = seq(min(x) - 1, max(x) + 1, length = 11)
> > > b
> >  [1] -3.4038769 -2.7451072 -2.0863375 -1.4275678 -0.7687980 -0.1100283
> >  [7]  0.5487414  1.2075111  1.8662808  2.5250506  3.1838203
> > >
> > > invisible(hist(x, breaks = b, include.lowest = TRUE, plot = FALSE))
> > Warning message:
> > argument 'include.lowest' is not made use of in: hist.default(x,
> > breaks = b, include.lowest = TRUE, plot = FALSE)
> >
> >
> > I don't think a warning is appropriate, since ?hist says:
> >
> > include.lowest: logical; if 'TRUE', an 'x[i]' equal to the 'breaks'
> >           value will be included in the first (or last, for 'right =
> >           FALSE') bar.  This will be ignored (with a warning) unless
> >           'breaks' is a vector.
> >
> > and in this case 'breaks' does qualify as a "vector" by my understanding.
> >
> > Note that the warning goes away with 'plot=TRUE'. This suggests that
> > this might have something to do with this vaguely worded entry in the
> > NEWS for R 2.4.0:
> >
> >    o   hist(*, plot = FALSE) now warns about unused arguments.
> >
> > Neither the help page nor the NEWS file indicates what arguments are
> > considered 'unused' when 'plot = FALSE', but either possibility ---
> > (1) include.lowest is actually unused and (2) it is used and the
> > warning is wrong --- is a bug.
>
> As I read the code, the purpose is to warn people if they supply plot
> arguments (density, xlim, ylim,....) while plot=FALSE.

That's not exactly what the NEWS entry says though.

> There's a stop
> list coded by
>
>         nf <- nf[is.na(match(nf, c("x", "breaks", "freq", "nclass",
>             "plot", "probability")))]
>
> and I think the problem is just that "include.lowest" should have been
> on the list.

Yes. And while you are at it, my suggestion would be to either take
'probability' and 'freq' off that list (which would make the NEWS
entry slightly more true) and/or clarify the intent of the warning.
The current help page is silent about the warnings when 'plot =
FALSE', and I don't think that's a good thing. It's not exactly a bug,
so I didn't file a bug report, but I did send a message to r-devel
about it a while back.

-Deepayan