Strange behavior of pretty
On Sat, 3 Nov 2007, Duncan Murdoch wrote:
On 03/11/2007 4:34 PM, Ulrike Gr?mping wrote:
Dear all,
I got stuck with a weird behavior of pretty:
pretty(c(-1,1),n=1)
yields -2, 0, 2 instead of the desired result -1 1.
Likewise,
pretty(c(-2,2),n=1)
yields -5, 0, 5 instead of the desired result -2 2.
I don't think pretty() should ever return -1 1 or -2 2. It should always return multiples of some round number, so it would have to include 0. But I don't see why it is expanding the range. It doesn't do that when the results are non-negative, e.g.
pretty(0:1, n=1)
[1] 0 1
pretty(0:2, n=1)
[1] 0 2
With n=0, the result is the same, while with n=2 everything is as documented. This behavior nastily messed up some of my cube plots. Is there any chance that this can be fixed?
The documentation has been sitting for 8 years with "FIXME" notes in it, so I wouldn't hold my breath.
The way to make the 'chance' appreciable would be to contribute a (high-quality) patch.
Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595