[Rspatial-devel] sp: SpatialGrid now directly inherits from Spatial (only)

[Roger Bivand]

On Mon, 19 Sep 2011, Roger Bivand wrote:

> On Sun, 18 Sep 2011, Edzer Pebesma wrote:
>
>> I committed a largish edit to sp today.
>> 
>> In particular, following my investigations reported here:
>> https://stat.ethz.ch/pipermail/r-devel/2011-September/062037.html
>> having SpatialGrid directly inheriting from SpatialPixels does not work,
>> as it needs an explicit coerce (coordinates are not there!), and setting
>> this coerce with setAs() does lead to unexpected results (essentially,
>> objects becoming "Spatial").
>> 
>> Now, SpatialGrid inherits from Spatial, and an explicit setIs() makes it
>> SpatialPixels too, and sets the right coerce function.
>> 
>> The following methods were moved from functions to S4 generics:
>> proj4string, proj4string<-, gridded, gridded<-, is.projected.
>> 
>
> Edzer,
>
> I see a failure in Ch 3 with the new code:
>
>> ###################################################
>> ### chunk number 11:
>> ###################################################
>> data(meuse.grid .... [TRUNCATED]
>
>> coordinates(meuse.grid) <- c("x", "y")
>
>> meuse.grid <- as(meuse.grid, "SpatialPixels")
>
>> image(meuse.grid, col = "grey")
> Error in image.default(meuse.grid, col = "grey") :
>  argument must be matrix-like
>
>
> And one in in Ch 4 that I made myself. Both could be related to rev 1139 in 
> image.R:
>
> https://r-forge.r-project.org/scm/viewvc.php/pkg/sp/R/image.R?root=rspatial&r1=1026&r2=1139

Only the Ch 4 error was from my revision.

The code is on p. 59 for Fig 3.1, the problem is that image() gets a 
SpatialPixels object, and as there is no image method, it goes to 
image.default. I've added image.SpatialPixels() to image.R, added it to 
NAMESPACE and the help page, and now things work. I think that something 
was inheriting backwards before, wasn't it, and that your revisions 
blocked that route? Revisions committed.

Roger

>
> This was motivated by an off-list request to suppress the warning given when 
> image(, breaks= ) is used as breaks is not an argument in plot.window, which 
> gets the ... send through.
>
> I'll try to fix this myself, and test again on the book code.
>
> Roger
>
>
>

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no