R-3.6.1-1.build.sh

#! /bin/bash

PKG=R
VERSION=3.6.1
BUILD=1

# https://www.r-project.org/
# https://cran.r-project.org/mirrors.html
CRANMIRROR=https://packages.othr.de/cran/

PREFIX=/pkg/$PKG-$VERSION-$BUILD
if [ -n "$TESTING" ]; then PREFIX=/dev/shm/$PKG-$VERSION-$BUILD ; fi

# export GITHUB_PAT='5***b' 38 secret chars
. ./github_pat.source
if [ -z "$GITHUB_PAT" ] ; then
  echo "# Error: GITHUB_PAT is not set, build of 'dynverse/dyno' is supposed to fail."
  echo "#        Check the dvteam mails for 'personal access token'"
  exit 1
fi

# neither home nor /usr/local/package - but texlive kpsewhich for nghiavtr/BPSC
PATH=/bin:/usr/bin:/usr/sbin:/usr/local/bin:/package/texlive/2015/bin

set -xe
umask 022

# Use this when R is build & installed and the package list 'work in progress'
# PKGDEBUG=1

BUILD_TMPDIR=/dev/shm/$PKG-$VERSION-$BUILD.build.tmp
test -d $BUILD_TMPDIR && rm -r $BUILD_TMPDIR
mkdir -p $BUILD_TMPDIR/home
export TMPDIR=$BUILD_TMPDIR
export HOME=$BUILD_TMPDIR/home

exec </dev/null

mkdir -p $PREFIX
cat >$PREFIX/profile <<-EOF
	PATH=$PREFIX/bin:\$PATH
	PKG_CONFIG_PATH=$PREFIX/lib/pkgconfig\${PKG_CONFIG_PATH:+:\$PKG_CONFIG_PATH}
	LD_LIBRARY_PATH=$PREFIX/lib\${LD_LIBRARY_PATH:+:\$LD_LIBRARY_PATH}
	if [ -d $PREFIX/.compatlibs ]; then export LD_LIBRARY_PATH=$PREFIX/.compatlibs\${LD_LIBRARY_PATH:+:\$LD_LIBRARY_PATH} ; fi
	export PKG_CONFIG_PATH LD_LIBRARY_PATH
EOF
. $PREFIX/profile

NPROC=$(nproc)
export MAKEFLAGS="-j $NPROC"

BUILDDIR=$BUILD_TMPDIR/build
mkdir -p $BUILDDIR
cd $BUILDDIR

if [ -z "$PKGDEBUG" ]; then

# Start with some extra stuff. Since this build includes a full
# installation of fairly recent boost templates (1.69), there is
# no argument against reinstalling gdal, proj, and the like (cf.
# /package/gis).

( # for 'dynverse/dyno', see UDUNITS2_INCLUDE, UDUNITS2_LIBS below
  UNITSVER=2.2.26
  test -e udunits-$UNITSVER.tar.gz || wget --no-verbose ftp://ftp.unidata.ucar.edu/pub/udunits/udunits-$UNITSVER.tar.gz
  test -d udunits-$UNITSVER        || tar -xf udunits-$UNITSVER.tar.gz
  cd udunits-$UNITSVER
  CFLAGS='-O2 -fPIC' \
  ./configure --prefix=$PREFIX --libdir=$PREFIX/lib --enable-shared=no
  make
  make install
)

# The rest is for 'sf', with 'spdep' as the real culprit.
# sf is a pain in the ass, as it carries tons of system dependencies (GDAL, PROJ, ...)
# https://github.com/r-spatial/sf - Simple Features for R
# ('Simple' in the sense of brainless I guess)
# OTOH its needed by monocle3, which is related to AG Mundlos & M. Spielmann, a Nature publication ...

(
  PROJVER=6.1.0
  test -e proj-$PROJVER.tar.gz || wget --no-verbose http://download.osgeo.org/proj/proj-$PROJVER.tar.gz
  test -d proj-$PROJVER || tar -xf proj-$PROJVER.tar.gz
  cd proj-$PROJVER

  ./configure \
  --prefix=$PREFIX \
  --libdir=$PREFIX/lib \
  --with-external-gtest=no

  make
  make install
)

(
  GEOSVER=3.7.2
  test -e geos-$GEOSVER.tar.bz2 || wget --no-verbose http://download.osgeo.org/geos/geos-$GEOSVER.tar.bz2
  test -d geos-$GEOSVER || tar -xf geos-$GEOSVER.tar.bz2
  cd geos-$GEOSVER

  ./configure \
  --prefix=$PREFIX \
  --libdir=$PREFIX/lib

  make
  make install
)

(
  GDALVER=2.4.1
  test -e gdal-$GDALVER.tar.xz || wget --no-verbose http://download.osgeo.org/gdal/$GDALVER/gdal-$GDALVER.tar.xz
  test -d gdal-$GDALVER || tar -xf gdal-$GDALVER.tar.xz
  cd gdal-$GDALVER

  export CFLAGS="-O2 -fPIC"
  export CXXFLAGS="-O2 -fPIC"
  ./configure \
  --prefix=$PREFIX \
  --libdir=$PREFIX/lib

  make
  make install
)

# ###########
# Now Build R

test -e R-$VERSION.tar.gz || wget --no-verbose https://cran.r-project.org/src/base/R-3/R-$VERSION.tar.gz
test -d R-$VERSION        || tar -xf R-$VERSION.tar.gz
cd R-$VERSION

sed -i 's/en_GB/en_US/g' tests/reg-tests-3.R tests/reg-tests-3.Rout.save

./configure \
  --prefix=$PREFIX \
  --libdir=$PREFIX/lib \
  --with-pic \
  --with-recommended-packages \
  --enable-shared \
  --with-x \
  --enable-R-shlib \
  LDFLAGS=-Wl,-rpath,$PREFIX/lib/R/lib

export TZ=CEST  # some tests need timezone
make
make check
make install

cd $BUILDDIR

cat <<- RCODEXXX | Rscript -
	options(repos=structure(c(CRAN='$CRANMIRROR')))
	install.packages('BiocManager')
	install.packages('remotes') # or installs from github will fail.
	BiocManager::install() # this should update the 'base' packages
RCODEXXX

echo "# State of lib/R/library after initial setup:"
ls -la $PREFIX/lib/R/library

fi ### -z "$PKGDEBUG"

unset  DISPLAY # avoid warnings about 'RGL: unable to open X11 display'
export UDUNITS2_INCLUDE=$PREFIX/include
export UDUNITS2_LIBS=$PREFIX/lib

# List of packages
# Package names containing a '/' are treated as GitHub repositories
#
# Format hint:
# cat package_list | sed -e 's/\s\s*/\n/g' | sort -f | uniq | tr '\n' ' ' | fmt -w 120

cat <<- RCODEXXX > package_list
 abind acepack ACME ade4 AdequacyModel affy affyio affyPLM akima ALL annotate AnnotationDbi AnnotationForge aod
 ape apeglm arrayQualityMetrics ash assertthat bamsignals base64 base64enc batchelor BatchJobs BayesDA BBmisc
 bdsmatrix beadarray BeadDataPackR beanplot beeswarm bgmm BH bigmemory bigmemory.sri Biobase BiocGenerics
 BiocParallel biomaRt BioNet Biostrings biovizBase bit bitops boot bootstrap Boruta bpca BradleyTerry2 brew
 brglm BSgenome BSgenome.Ecoli.NCBI.20080805 bumphunter c060 C50 Cairo car caret Category caTools CellNOptR
 checkmate ChIPpeakAnno chipseq chron circlize CircStats Ckmeans.1d.dp class clue cluster CNEr coda codetools coin
 cole-trapnell-lab/leidenbase cole-trapnell-lab/monocle3 colorspace combinat corpcor corrgram corrplot crayon
 cummeRbund curl data.table DBI DelayedArray DelayedMatrixStats deldir DEoptimR DESeq DESeq2 devtools DEXSeq
 dichromat digest DirichletMultinomial dismo DNAcopy doMC doParallel doRNG doSNOW dostats dplyr dtw dynamicTreeCut
 dynverse/dyno e1071 earth easyRNASeq edgeR elasticnet energy entropy evaluate evd exactRankTests exomeCopy
 ExomeDepth fail fANCOVA fastcluster fastICA fastmatch FDb.InfiniumMethylation.hg19 fdrtool FField fields filehash
 flashClust foreach foreign formatR Formula futile.logger futile.options gbm gclus gcrma gdata genefilter GeneNet
 geneplotter genetics GenomeGraphs GenomeInfoDb genomeIntervals GenomicAlignments GenomicFeatures GenomicRanges
 GEOquery getopt GGally ggbio ggm ggplot2 git2r glasso Glimma glmnet GlobalAncova GlobalOptions globaltest gmp
 gnm GO.db GOstats GOTHiC gplots graph gRbase gridBase gridExtra gridSVG GSEABase GSEAlm gsl gsubfn gtable gtools
 Gviz h2o haplo.stats HardyWeinberg hash hdf5r hdi hdrcde hexbin hflights hgu95av2 hgu95av2.db hgu95av2cdf highr
 Hmisc HMM Homo.sapiens hsmm htmltools HTqPCR httpuv httr hwriter igraph illuminaio impute inline intervals ipred
 IRanges iterators itertools jokergoo/ComplexHeatmap jokergoo/EnrichedHeatmap jsonlite kernlab KernSmooth knitr
 ks labeling Lahman lambda.r lars lattice latticeExtra lava lazyeval leaps LearnBayes lhs limma linprog lintr lme4
 lmtest locfit longitudinal lpSolve LSD lumi magrittr manipulate mapproj maps maptools maptree markdown MASS MAST
 Matrix MatrixEQTL matrixStats mclust mda MEDIPS memoise meta MethylSeekR methylumi mgcv mice microbenchmark mime
 minfi minqa misc3d mitools mixtools mlegp mnormt modeltools monocle multcomp multicool multtest munsell muscle
 mvtnorm nghiavtr/BPSC nleqslv nlme nloptr nnet nor1mix numDeriv OmicCircos optparse org.Hs.eg.db OrganismDbi pamr
 party partykit pbapply pbkrtest pcalg pcaMethods penalizedSVM peperr PerfMeas pheatmap pkgmaker plotmo plotrix pls
 plus plyr preprocessCore pROC prodlim profileModel proto proxy pwr qsea qtl quadprog quantreg QuasR qvalue qvcalc
 R.cache R.methodsS3 R.oo R.utils R2HTML R6 randomForest randomGLM raster rbenchmark RBGL RCircos RColorBrewer Rcpp
 RcppArmadillo RcppEigen RCurl registry relaimpo relimp remotes reshape reshape2 reticulate rex rFerns rggobi rgl
 Rgraphviz RGtk2 rhdf5 RhpcBLASctl Rhtslib rJava rjson RJSONIO rlecuyer rmarkdown rmeta RMySQL RNAseq123 rnaseqGene
 rngtools robustbase ROC ROCR roxygen2 rpart RRF Rsamtools RSQLite rstudioapi rtracklayer RUnit rversions RWeka
 RWekajars S4Vectors sandwich scales scalreg scatterplot3d scDD SCnorm scone SCORPIUS sda segmented sendmailR seqinr
 seqLogo seriation setRNG Seurat sf sfsmisc shape shiny ShortRead siggenes simpleaffy SingleCellExperiment sm snow
 snowfall snpStats som sp spam SparseM spatial splancs spls statmod stepPlr stringdist stringi stringr strucchange
 SummarizedExperiment SuppDists survey survival svMisc svTools tables TeachingDemos testthat TFBSTools TFMPvalue tgp
 TH.data tidyr topGO TSP TxDb.Hsapiens.UCSC.hg19.knownGene VariantAnnotation vcd VennDiagram venneuler verification
 VGAM vsn waveslim WGCNA whisker xlsx xlsxjars XML xml2 xtable XVector yaml yeastCC zinbwave zipfR zlibbioc zoo
RCODEXXX

cat <<- RCODEXXX | Rscript -
  options(repos=structure(c(CRAN='$CRANMIRROR')))
  pkgs <- scan('package_list', what = character())
  for (pk in pkgs) {
    pa <- unlist(strsplit(pk, '/')); p <- pa[1]; if (! is.na(pa[2])) p <- pa[2]
    # there is no simple way to get rid of the error message, so avoid mimimi...
    cat('* if there is no package, install it - errors here are part of the test!\n')
    chk <- NULL; try(chk <- find.package(p))
    if (is.null(chk)) {
      BiocManager::install(pk, Ncpus = $NPROC, update = FALSE) # we are fresh within an 2h interval
      ret <- library(p, character.only = TRUE, logical.return = TRUE)
      if (! ret) q(save = "no", status = 2)
      detach( paste("package", p, sep = ":"), unload = TRUE, character.only=TRUE)
      cat( sprintf('* REALY_DONE with: %s\n', pk))
    } else {
      cat( sprintf('* ALREADY_DONE with: %s\n', pk))
    }
  }
RCODEXXX

# simple test: can everything be found?
cat <<- RCODEXXX | Rscript -
  pkgs <- scan('package_list', what = character())
  for (pk in pkgs) {
    pa <- unlist(strsplit(pk, '/')); p <- pa[1]; if (! is.na(pa[2])) p <- pa[2]
    chk <- NULL; try(chk <- find.package(p))
    if (is.null(chk)) {
      cat( sprintf('* ERROR: Missing package: %s\n', pk))
      q(save = "no", status = 3)
    }
  }
RCODEXXX

# stress test: load all packages (uses 4G, claims 16G)
cat <<- RCODEXXX | Rscript -
  pkgs <- scan('package_list', what = character())
  for (pk in pkgs) {
    pa <- unlist(strsplit(pk, '/')); p <- pa[1]; if (! is.na(pa[2])) p <- pa[2]
    ret <- library(p, character.only = TRUE, logical.return = TRUE)
    if (! ret) {
      cat( sprintf('* ERROR: Failed to load package/library: %s\n', pk))
      q(save = "no", status = 4)
    }
  }
RCODEXXX

# rstudios rsession does a lot of magic to detect the true R binary, thus bypassing
# our friendly 'hints' from the wrapper.
# tell lib/R/library/sf/libs/sf.so about lib -- yes it's the beloved sf package again!
# the original rpath is only set to /pkg/R-3.6.1-0/lib/R/lib, add /pkg/R-3.6.1-0/lib.
# to be fixed in next rstudio build

patchelf --set-rpath $PREFIX/lib/R/lib:$PREFIX/lib \
         $PREFIX/lib/R/library/sf/libs/sf.so

exit

# NOTES:
#
# From git to bioconductor:
#   YosefLab/scone       # 29.03.2017 pacini  1.3.0 -> 1.8.0
#   rhondabacher/SCnorm  # 29.03.2017 pacini  1.5.7 -> 1.6.0
#
# Goners (not available for R version 3.6.x):
#   BiocInstaller Slingshot bigrf grofit harvestr
#   lint setwidth spdepbio survrec