R-4.4.1-0.build.sh

#! /usr/bin/bash

# COOKIE=$(mcookie|cut -c-8); grep -v V_GREP_ME $0 > /dev/shm/runme-$COOKIE.sh ; sleep 0.3; exec bash /dev/shm/runme-$COOKIE.sh
# TESTING=1

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

PKG=R
VERSION=4.4.1
BUILD=0

# https://www.r-project.org/
# https://cran.r-project.org/mirrors.html

# A caching proxy under /project/retarded/rproxy
CRANMIRROR=http://platsch:8010/cran/$PKG-$VERSION-$BUILD/cran
BIOCMIRROR=http://platsch:8010/bioconductor/$PKG-$VERSION-$BUILD/bioc
GITMIRROR=http://platsch:8010/gitclone/$PKG-$VERSION-$BUILD

PREFIX=/pkg/$PKG-$VERSION-$BUILD
if [ -n "$TESTING" ]; then PREFIX=/scratch/local2/$PKG-$VERSION-$BUILD ; fi

echo "# Ready player one."
date

set -vxe
umask 022

BUILD_TMPDIR=/scratch/local2/$PKG-$VERSION-$BUILD-$USER.build.tmp
test -d $BUILD_TMPDIR && rm -fr $BUILD_TMPDIR/build/*
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
source $PREFIX/profile


NPROC=$(( $(nproc) * 4 / 5 + 1 ))
export MAKEFLAGS="-j $NPROC"

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

if [ -z "$PKGDEBUG" ]; then

# Start with some extra stuff. Mostly owed to the inability to keep dependencies under control :/

( # for 'dynverse/dyno', see UDUNITS2_INCLUDE, UDUNITS2_LIBS below
  UNITSVER=2.2.28
  #  Orig:https://downloads.unidata.ucar.edu/udunits/2.2.28/udunits-2.2.28.tar.gz
  BEEHIVE=https://beehive.molgen.mpg.de/58259d94f766c13b5b0cf1aed92ebbe3/udunits-2.2.28.tar.gz
  test -e udunits-$UNITSVER.tar.gz || wget -nv $BEEHIVE
  test -d udunits-$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
)

(
  PROJVER=9.2.0
  #  Orig: http://download.osgeo.org/proj/proj-9.2.0.tar.gz
  BEEHIVE=https://beehive.molgen.mpg.de/1241c7115d8c380ea19469ba0828a22a/proj-9.2.0.tar.gz
  test -e proj-$PROJVER.tar.gz || wget -nv $BEEHIVE
  test -d proj-$PROJVER || tar -xf proj-$PROJVER.tar.gz
  cd proj-$PROJVER

  mkdir -p build; cd build

  cmake \
    -DCMAKE_INSTALL_PREFIX=$PREFIX \
    -DCMAKE_BUILD_TYPE=RelWithDebInfo \
    -DBUILD_TESTING=OFF \
    -DBUILD_BENCHMARKS=OFF \
    -DRUN_NETWORK_DEPENDENT_TESTS=OFF \
    ..

  make
  make install
)

(
  GEOSVER=3.11.2
  #  Orig: http://download.osgeo.org/geos/geos-3.11.2.tar.bz2
  BEEHIVE=https://beehive.molgen.mpg.de/5967d369c641f0fad42258fcb22af6b5/geos-3.11.2.tar.bz2
  test -e geos-$GEOSVER.tar.bz2 || wget -nv $BEEHIVE
  test -d geos-$GEOSVER || tar -xf geos-$GEOSVER.tar.bz2
  cd geos-$GEOSVER

  mkdir -p build; cd build

  cmake \
    -DCMAKE_INSTALL_PREFIX=$PREFIX \
    -DCMAKE_BUILD_TYPE=RelWithDebInfo \
    -DBUILD_TESTING=OFF \
    ..

  make
  make install
)

(
  GDALVER=3.6.4
  #  Orig: http://download.osgeo.org/gdal/3.6.4/gdal-3.6.4.tar.gz
  BEEHIVE=https://beehive.molgen.mpg.de/7128d0309f9fe2266d72ea7228a3817f/gdal-3.6.4.tar.gz
  test -e gdal-$GDALVER.tar.gz || wget -nv $BEEHIVE
  test -d gdal-$GDALVER || tar -xf gdal-$GDALVER.tar.gz
  cd gdal-$GDALVER

  mkdir -p build; cd build

  # still required for building... (and why use -DBUILD_CSHARP_BINDINGS and -DBUILD_JAVA_BINDINGS ???)
  source /pkg/mono-6.10.0-0/profile
  source /pkg/openjdk-11.0.3.2-0/profile

  cmake \
    -DCMAKE_INSTALL_PREFIX=$PREFIX \
    -DCMAKE_BUILD_TYPE=RelWithDebInfo \
    -DBUILD_CSHARP_BINDINGS=OFF \
    -DBUILD_JAVA_BINDINGS=OFF \
    -DBUILD_PYTHON_BINDINGS=OFF \
    -DBUILD_TESTING=OFF \
    -DGDAL_BUILD_OPTIONAL_DRIVERS=OFF \
    -DOGR_BUILD_OPTIONAL_DRIVERS=OFF \
    ..

  make
  make install
)

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

#  Orig:https://cran.r-project.org/src/base/R-4/R-4.4.1.tar.gz
BEEHIVE=https://beehive.molgen.mpg.de/cfe520ea9fbca4f3c3d8462aaee7af46/R-4.4.1.tar.gz
test -e R-$VERSION.tar.gz || wget -nv $BEEHIVE
test -d R-$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

# set R_LD_LIBRARY_PATH for etc/ldpaths, so R gets the correct
# env if it is started without sourcing the profile
R_LD_LIBRARY_PATH=${PREFIX}/lib:${PREFIX}/.compatlibs \
./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='$CRANMIRROR')))
	options(BioC_mirror = '$BIOCMIRROR')
	install.packages('BiocManager')
	install.packages('remotes')
	install.packages('devtools')
	BiocManager::install(update = TRUE, ask = FALSE) # update the 'base' packages
RCODEXXX

# java stuff seems to be a troublemaker, install first
cat <<- RCODEXXX | Rscript -
	options(repos=structure(c(CRAN='$CRANMIRROR')))
	options(BioC_mirror = '$BIOCMIRROR')
	BiocManager::install(c('rJava', 'RWeka', 'RWekajars'), update = FALSE)
RCODEXXX

# same applies for V8 (JavaScript, Google)
(
  export DOWNLOAD_STATIC_LIBV8=1
  cat <<- RCODEXXX | Rscript -
		options(repos=structure(c(CRAN='$CRANMIRROR')))
		install.packages('V8')
	RCODEXXX
)

# apaches 'arrow' library is said to be hard to build under linux,
# so use the binary package from rstudio (now 'posit')
# https://arrow.apache.org/docs/r/articles/install.html
(
  cat <<- RCODEXXX | Rscript -
	options( HTTPUserAgent = sprintf( "R/%s R (%s)", getRversion(),
			paste( getRversion(), R.version["platform"], R.version["arch"], R.version["os"] )
		)
	)
	install.packages("arrow", repos = "https://packagemanager.rstudio.com/all/__linux__/focal/latest")
	RCODEXXX
)

# ncdf4 (netcdf binding), suffers a bit from our statically build netcdf, which requires some basic libs itself
(
  #orig: https://cran.r-project.org/src/contrib/ncdf4_1.22.tar.gz
  BEEHIVE=https://beehive.molgen.mpg.de/cdfddec28e78da74ed216052d469a585/ncdf4_1.22.tar.gz
  test -e ncdf4_1.22.tar.gz || wget -nv $BEEHIVE
  test -d ncdf4_1.22        || tar -xf ncdf4_1.22.tar.gz
  cd ncdf4 # not ncdf4_1.22 !
  export LDFLAGS="$(nc-config --all | grep -e '--static' | cut -d'>' -f2)"
  R CMD INSTALL -d .
)

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

# 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 AUCell 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 colorspace combinat corpcor corrgram corrplot cowplot crayon cummeRbund curl cvTools data.table DBI
 DelayedArray DelayedMatrixStats deldir DEoptimR DESeq2 deSolve DEXSeq dichromat digest DirichletMultinomial
 DNAcopy doMC doParallel doRNG doSNOW dostats dplyr dtw dynamicTreeCut e1071 earth EBImage edgeR elasticnet
 energy entropy evaluate evd exactRankTests ExomeDepth ExperimentHub fail fANCOVA fastcluster fastICA fastmatch
 FDb.InfiniumMethylation.hg19 fdrtool fields filehash flashClust foreach foreign formatR Formula futile.logger
 futile.options gbm gclus gcrma gdata genefilter GeneNet geneplotter genetics GenomeInfoDb genomeIntervals
 GenomicAlignments GenomicFeatures GenomicRanges GEOquery getopt GGally ggbio ggExtra ggformula ggm
 ggplot2 ggpubr ggstatsplot 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
 HiCcompare highr HiveR Hmisc HMM Homo.sapiens htmltools httpuv httr hwriter igraph illuminaio imcRtools impute
 infotheo inline intervals ipred IRanges IRkernel iterators itertools jsonlite kernlab KernSmooth knitr ks
 kSamples labeling Lahman lambda.r lars lattice latticeExtra lava lazyeval leaps LearnBayes lhs limma linprog
 lintr lme4 lmtest locfit longitudinal lpSolve lpSolveAPI LSD lumi magrittr manipulate mapproj maps maptree
 markdown MASS MAST Matrix MatrixEQTL matrixStats mclust mda MEDIPS memoise meta MethylSeekR methylumi mgcv
 mice microbenchmark mime minfi minqa misc3d mitools mixtools mlegp modeltools multcomp multicool
 multiHiCcompare multtest munsell muscle mvtnorm mzR nleqslv nlme nloptr NMF nnet nor1mix numDeriv OmicCircos
 optparse org.Hs.eg.db OrganismDbi pamr party partykit pbapply pbkrtest pcalg pcaMethods penalizedSVM peperr
 pheatmap PhyloProfile pkgmaker plotmo plotrix pls plyr ppcor pracma preprocessCore pROC prodlim profileModel
 proto proxy pwr qsea qtl quadprog quantreg QuasR qvalue qvcalc R.cache R.matlab R.methodsS3 R.oo R.utils
 R2HTML R6 RaMS randomForest randomGLM rbenchmark RBGL RCircos RcisTarget RColorBrewer Rcpp RcppArmadillo
 RcppEigen RCurl readxl registry relaimpo relimp remotes reshape reshape2 reticulate rex rFerns rgl Rgraphviz
 rhandsontable rhdf5 RhpcBLASctl Rhtslib rJava rjson RJSONIO rlecuyer RMariaDB rmarkdown rmeta Rmpfr RMySQL
 RNAseq123 rnaseqGene rngtools robustbase ROC ROCR roxygen2 rpart RPostgres RPostgreSQL RRF Rsamtools
 RSQLite rstudioapi rtracklayer RUnit rversions S4Vectors sandwich scales scalreg scatterplot3d SCnorm
 scone SCORPIUS sda segmented sendmailR seqinr seqLogo seriation setRNG Seurat sf shape shiny shinythemes
 ShortRead siggenes SingleCellExperiment sm snow snowfall snpStats som sp spam SparseM spatial splancs spls
 statmod stepPlr stringdist stringi stringr strucchange SummarizedExperiment SuppDists survey svMisc svTools
 tables TeachingDemos terra testthat TFBSTools TFMPvalue tgp TH.data tidyr tidyverse tiff topGO tradeSeq
 TSP TxDb.Hsapiens.UCSC.hg19.knownGene VariantAnnotation vcd VennDiagram venneuler verification VGAM vsn
 waveslim WGCNA whisker writexl xlsx xlsxjars XML xml2 xtable XVector yaml yeastCC zinbwave zipfR zlibbioc zoo
RCODEXXX

cat <<- RCODEXXX | Rscript -
  XSAVE_PATH <- Sys.getenv('PATH')
  options(repos=structure(c(CRAN='$CRANMIRROR')))
  options(BioC_mirror = '$BIOCMIRROR')
  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)) {
      cat( sprintf('* About to install: %s\n', pk), file='work_monitor_list', append = TRUE)
      BiocManager::install(pk, update = FALSE)
      ret <- library(p, character.only = TRUE, logical.return = TRUE)
      if (! ret) q(save = "no", status = 2)
      # the logic fails for 'cole-trapnell-lab/cicero-release', fixed below
      detach( paste("package", p, sep = ":"), unload = TRUE, character.only=TRUE)
      if (Sys.getenv('PATH') != XSAVE_PATH) {
        cat( sprintf('*** PATH SCREWUP OCCURED (package %s)\n', pk))
        cat( sprintf('*** expected: %s\n', XSAVE_PATH))
        cat( sprintf('*** now:      %s\n', Sys.getenv('PATH')))
      }
      Sys.setenv(PATH = XSAVE_PATH)
      cat( sprintf('* REALY_DONE with: %s\n', pk))
    } else {
      cat( sprintf('* ALREADY_DONE with: %s\n', pk))
    }
  }
RCODEXXX

( # github nghiavtr/BPSC
  mkdir -p BPSC
  cd BPSC
  # --trust-server-names makes wget use the 'redirected' name
  wget -nv --trust-server-names platsch $GITMIRROR/nghiavtr/BPSC && test -e *.latest.tar.gz
  tar -xf *.latest.tar.gz
  cd BPSC
  cat <<- RCODEXXX | Rscript -
		devtools::install_local('.')
	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

# do the remaining git things.

# do 'cole-trapnell-lab/cicero-release' seperately
echo "BiocManager::install('cole-trapnell-lab/cicero-release', update = FALSE)" | Rscript -

# same with 'GreenleafLab/ArchR', https://www.archrproject.com/index.html
cat <<- RCODEXXX | Rscript -
	options(repos=structure(c(CRAN='$CRANMIRROR')))
	options(BioC_mirror = '$BIOCMIRROR')
	devtools::install_github("GreenleafLab/ArchR", ref="master", repos = BiocManager::repositories())
	library(ArchR)
	ArchR::installExtraPackages()
RCODEXXX

# support for handling DiaNN output (https://github.com/vdemichev/DiaNN)
cat <<- RCODEXXX | Rscript -
	options(repos=structure(c(CRAN='$CRANMIRROR')))
	options(BioC_mirror = '$BIOCMIRROR')
	devtools::install_github("https://github.com/vdemichev/diann-rpackage", upgrade="never")
RCODEXXX

# finally a source build, this is/was intended to provide a clean velocyto.R package.
# velocyto.R uses an inappropriate makefile and has a silly multitreading issue

cd $BUILDDIR
mkdir -p velocyto_fixup
cd velocyto_fixup

git clone https://github.com/velocyto-team/velocyto.R
cd velocyto.R
git checkout 0.6 --force
git clean -dffx

mv src/Makevars src/Makevars.weird
# yields just -fopenmp, which won't work on macs with clang
echo 'PKG_CXXFLAGS=-Wall $(SHLIB_OPENMP_CXXFLAGS)' > src/Makevars

# intercept the inane 'grab all cores' policy
# defaultNCores <- function() { parallel::detectCores(logical=F) }

cp -p R/momentum_routines.R R/momentum_routines.R.orig
sed -e '/^defaultNCores/ d' -i R/momentum_routines.R
cat <<__NCores_HACK >> R/momentum_routines.R
defaultNCores <- function() {
  env <- Sys.getenv('VELOCYTO_CORES')
  if (env != "") { return(as.integer(env)) }
  parallel::detectCores(logical=F)
}
__NCores_HACK

R CMD INSTALL .

echo "# Game over ..."
date

exit