Packages Recommendation for Ubuntu 1604

Hi All - Our team builds a specialty Linux distribution focused on geospatial data and methods, primarily representing OSGeo dot-org incubated projects. You can see the last edition at The current build is based on LUbuntu 1604, and will remain so for the lifecycle of that OS. This product will be distributed freely at FOSS4G 2017 Boston, and other events.

I am asking the rOpenSci community for recommendations on packages and versions for R on Ubuntu 1604. The last alpha build shows R 3.4.0 and the following contents. Suggestions? thanks in advance

ip <-[,c(1,3:4)])
rownames(ip) <- NULL
ip <- ip[$Priority),1:2,drop=FALSE]
print(ip, row.names=FALSE)
Package Version
IRdisplay 0.4.4
repr 0.12.0
abind 1.4-3
bitops 1.0-6
caTools 1.17.1
classInt 0.1-23
coda 0.18-1
colorspace 1.2-4
DCluster 0.2-7
deldir 0.1-9
e1071 1.6-7
FNN 1.1
gdata 2.17.0
geoR 1.7-5.2
gmodels 2.16.2
gplots 2.17.0
gstat 1.1-3
gtools 3.5.0
intervals 0.15.1
LearnBayes 2.15
littler 0.3.2
maptools 0.8-30
ncdf4 1.15
pkgKitten 0.1.3
plyr 1.8.3
polyclip 1.3-0
RandomFields 3.0.44
raster 2.5-8
RColorBrewer 1.1-2
Rcpp 0.12.3
RcppArmadillo 0.6.500.4.0
rgdal 1.1-10
rgeos 0.3-19
rgrass7 0.1-8
RSAGA 0.94-5
shapefiles 0.7
sp 1.2-3
spacetime 1.1-5
spam 1.0-1
spatstat 1.37-0
spdep 0.6-5
splancs 2.01-38
tcltk2 1.2-11
tensor 1.5
tkrplot 0.0-23
XML 3.98-1.3
xts 0.9-7
zoo 1.7-12

You may want to look at the packages installed on the rocker R geospatial Docker container: It has a relatively comprehensive set of important R geospatial packages and dependencies.

Here’s the list from the Dockerfile. A notable package not on your list: sf, which is rapidly becoming a replacement for sp.

  && install2.r --error \
    RColorBrewer \
    RandomFields \
    classInt \
    deldir \
    gstat \
    lidR \
    mapdata \
    maptools \
    mapview \
    proj4 \
    raster \
    rgdal \
    rgeos \
    rlas \
    sf \
    sp \
    spacetime \
    spatstat \
    spdep \
    geoR \
    geosphere \

thx - that is super-useful … in fact, since we are on a deadline, we have rebuilt the R setup now… Berkeley + Athens collaboration, in 2017 :wink: Here is the list at the moment:

classint - 0.1-24-1cran1ppa0
dcluster - 0.2-7-1cran2ppa2
dismo - 1.1-4-1cran1ppa1
fnn - 1.1-1cran2ppa2
gdalutils -
geor - 1.7-5.2-1cran2ppa2
geosphere - 1.5-5-1cran2ppa1
gstat - 1.1-5-1cran1ppa1
intervals - 0.15.1-1cran2ppa2
mapdata - 2.2-6-1cran2ppa2
mapview - 2.0.1-1cran1ppa0
ncdf4 - 1.16-1cran1ppa1
opencpn - 4.6.1-0~xenial1
proj4 - 1.0-8-1cran3ppa2
r-base - 3.4.0-1~xenial0
raster - 2.5-8-1cran2ppa1
rgdal - 1.2-7-1cran1ppa0
rgeos - 0.3-23-1cran1ppa1
rgrass7 - 0.1-9-1cran1ppa0
rsaga - 0.94-5-1cran1ppa0~ubuntu16.04.1~ppa1
sf - 0.5-0-1cran1~ubuntu16.04.1~ppa1
shapefiles - 0.7-1cran1precise3~ubuntu16.04.1~ppa1
sp - 1:1.2-3-1cran1ppa1
spacetime - 1.2-0-1cran1ppa0
spdep - 0.6-13-1cran1ppa0
splancs - 2.01-40-1cran1ppa0
zygrib - 8.0.1+dfsg.1-1~xenial1

@mdsumner any thoughts on this?

Here’s the ever-updating recipe I use:


WARNING - it’s not to be applied as an actual script in this form, because it’s hardcoded to a mirror I want, and with “a recent RStudio server specific version” - basically it’s a copy of the last install I did.

Some of the sf features will definitely need the up to date GDAL, and sf should definitely be highlighted by this project if possible.

In my experience if sf, rgdal, rgeos, dplyr, devtools and RNetCDF, ncd4 can be installed and used then everything else just installs easily. I’d like to see

tidyverse, devtools, RSQLite, ff, spatstat, rlas, lidR

added to the package list.

RStudio itself would be a very good addition too.


The sf travis configuration is another touchstone I constantly refer to:

1 Like

Just a note that rocker/geospatial is intended as a community effort. Please suggest additional packages that should be included in the the issues tracker: (Note that geospatial already inherits from tidyverse).

I’ve tried to include some basic pointers as to why the geospatial includes the packages it does in the README; I realize this can get subjective and welcome more input. Emphasis is naturally on those packages that are difficult to install otherwise (and favoring general packages over super-specialized ones).

quick update - the Beta 1 build of is due in about 24 hours from now… thanks to the excellent feedback here, R geospatial will be much improved.

I believe we are currently having some difficulty building cleanly, however. Many prerequisites are in place. Please contact the mailing list at for more details.

ps- I think people may realize by now, that it is in fact ppa:ubuntugis/ubuntugis-unstable that we are building, and that PPA will reflect some of the improvements discussed here.