Bug in R quartz() device

quartzbug.R

# I believe I found a bug in R's quartz() device, specifically it's
# GECap method (the C routine called by dev.capture).  The following
# yields the two images in the comment below.  This occurs whether working
# from the terminal or RStudio.  This works fine with Cairo devices that
# support raster capture.

quartz()
plot(1,1)
aa <- dev.capture()
dev.size("px")
#> [1] 504 504
dim(aa)
#> [1] 751 795
# Note that the size of the raster does not match the device size.
plot(as.raster(aa))  # Looks way different.
tail(table(aa))
#> #FFF9F9 #FFFBFB #FFFEFE     red    snow   white
#>      1       4       2    3420       1  589997
# Note all the red pixels in what should be a black and white image
dim(aa)

# I note I am on a retina Macbook.  I have XQuartz 2.7.8.
devtools::session_info()

# Session info ------------------------------------------------------------------
#   setting  value
# version  R version 3.3.2 (2016-10-31)
# system   x86_64, darwin13.4.0
# ui       X11
# language (EN)
# collate  en_US.UTF-8
# tz       America/New_York
# date     2016-12-10
# 
# Packages ----------------------------------------------------------------------
#   package    * version     date       source
# crayon     * 1.3.2       2016-06-28 cran (@1.3.2)
# devtools     1.12.0.9000 2016-12-08 Github (hadley/devtools@1ce84b0)
# digest       0.6.10      2016-08-02 cran (@0.6.10)
# memoise      1.0.0.9001  2016-12-04 Github (hadley/memoise@e392c7b)
# pkgbuild     0.0.0.9000  2016-11-28 Github (r-pkgs/pkgbuild@65eace0)
# pkgload      0.0.0.9000  2016-11-28 Github (r-pkgs/pkgload@def2b10)
# rstudioapi   0.6         2016-06-27 cran (@0.6)
# withr        1.0.2       2016-06-20 CRAN (R 3.3.0)

First Plot:

Second Plot:

I get the red graph too, not as ugly as Noam's (unless Noam's has been mangled in uploading); margins are about as different as one would expect from the reported difference between device size and capture size.

No, it wasn't mangled in upload, that's how it looks on my machine. I have a suspicion it has to do with the retina screen.

Thanks, the RGBA vs ARGB issue should now be fixed.