R devel - Nov 2017 - Height not set properly in grDevices::jpeg() with type = "cairo"

Hi,
I have been having issues producing plots in JPEG format, using type
"cairo" to get better anti-aliasing. When trying to set the physical
size with units = "cm" or units = "mm", the width is set
correctly but
the height is not - it looks like the height is simply treated as
pixels regardless of the 'units' argument.

Example:


x = 1:10
y = 2 * x
jpeg("ExamplePlot.jpg",
     type = "cairo",
     width = 200,
     height = 200,
     units = "mm",
     res = 96)
plot(x, y)
dev.off()


On my system (Windows 7, running R 3.4.2), this produces a plot that
is 755 x 200 pixels, and is vertically very squashed.

Looking at grDevices::jpeg, it looks like the culprit is these lines:

   g <- .geometry(width, height, units, res)
    if (match.arg(type) == "cairo") {
        antialias <- match(match.arg(antialias), aa.cairo)
        invisible(.External(C_devCairo, filename, 3L, g$width,
            height, pointsize, bg, res, antialias, quality, if
(nzchar(family)) family else "sans",
            300))
    }

g$width is used, but "height" is used instead of "g$height".
I suspect
simply using "g$height" here would fix the issue but have not had time
to test this.

My R.version:

platform       x86_64-w64-mingw32
arch           x86_64
os             mingw32
system         x86_64, mingw32
status
major          3
minor          4.2
year           2017
month          09
day            28
svn rev        73368
language       R
version.string R version 3.4.2 (2017-09-28)
nickname       Short Summer

Please let me know if any further information is needed.

Thanks,
Marius

Hi

That fix checks out for me so I have committed the change to r-devel.

Thanks for the report!

Paul

On 28/11/17 16:25, Marius wrote:
> Hi,
> I have been having issues producing plots in JPEG format, using type >
"cairo" to get better anti-aliasing. When trying to set the physical
> size with units = "cm" or units = "mm", the width is
set correctly but
> the height is not - it looks like the height is simply treated as
> pixels regardless of the 'units' argument.
> 
> Example:
> 
> 
> x = 1:10
> y = 2 * x
> jpeg("ExamplePlot.jpg",
>       type = "cairo",
>       width = 200,
>       height = 200,
>       units = "mm",
>       res = 96)
> plot(x, y)
> dev.off()
> 
> 
> On my system (Windows 7, running R 3.4.2), this produces a plot that
> is 755 x 200 pixels, and is vertically very squashed.
> 
> Looking at grDevices::jpeg, it looks like the culprit is these lines:
> 
>     g <- .geometry(width, height, units, res)
>      if (match.arg(type) == "cairo") {
>          antialias <- match(match.arg(antialias), aa.cairo)
>          invisible(.External(C_devCairo, filename, 3L, g$width,
>              height, pointsize, bg, res, antialias, quality, if
> (nzchar(family)) family else "sans",
>              300))
>      }
> 
> g$width is used, but "height" is used instead of
"g$height". I suspect
> simply using "g$height" here would fix the issue but have not had
time
> to test this.
> 
> My R.version:
> 
> platform       x86_64-w64-mingw32
> arch           x86_64
> os             mingw32
> system         x86_64, mingw32
> status
> major          3
> minor          4.2
> year           2017
> month          09
> day            28
> svn rev        73368
> language       R
> version.string R version 3.4.2 (2017-09-28)
> nickname       Short Summer
> 
> Please let me know if any further information is needed.
> 
> Thanks,
> Marius
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
-- 
Dr Paul Murrell
Department of Statistics
The University of Auckland
Private Bag 92019
Auckland
New Zealand
64 9 3737599 x85392
paul at stat.auckland.ac.nz
http://www.stat.auckland.ac.nz/~paul/