R-4.2.2-0.build.sh

#! /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=4.2.2
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

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

echo "# Ready player one."
date

set -xe
umask 022

BUILD_TMPDIR=/scratch/local2/$PKG-$VERSION-$BUILD-$USER.build.tmp
test -d $BUILD_TMPDIR && rm -fr $BUILD_TMPDIR/build/velocyto_fixup
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) * 4 / 5 + 1 ))
export MAKEFLAGS="-j $NPROC"

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

if [ -z "$PKGDEBUG" ]; then

( # seurat likes to have this one. Sigh, meaning some GIS-stuff remains in the bio-R ...
  GEOSVER=3.10.3
  # ORIG: https://download.osgeo.org/geos/geos-3.10.3.tar.bz2
  BEEHIVE=https://beehive.molgen.mpg.de/a5a6959a8a294d9dab778a9ef03ccced/geos-3.10.3.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 \
    -DBUILD_TESTING=OFF \
    ..

  make
  make install
)

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

# test -e R-$VERSION.tar.gz || wget --no-verbose https://cran.r-project.org/src/base/R-4/R-4.2.1.tar.gz
BEEHIVE=https://beehive.molgen.mpg.de/b154500667b5ebb480f1086dd817f017/R-4.2.2.tar.gz
test -e R-$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 LD_LIBRARY_PATH for etc/ldpaths, so R get the correct env
# if it is start without sourcing of 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') # or installs from github will fail.
	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

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'

# List of packages
# Package names containing a '/' are treated as GitHub repositories
# SingleCellExperiment must come before aertslab/SCENIC, so just do all github stuff at the end

# Format hint:
# cat package_list | sed -e 's/\s\s*/\n/g' | sort -f | uniq | tr '\n' ' ' | fmt -w 120

# ends with 805 dirs in /dev/shm/R-4.2.1-0/lib/R/library

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 ChIPseeker chipseq chron circlize CircStats Ckmeans.1d.dp class clue cluster clusterProfiler CNEr coda
 codetools coin colorspace combinat corpcor corrgram corrplot cowplot crayon cummeRbund curl cvTools data.table DBI
 DelayedArray DelayedMatrixStats deldir DEoptimR DESeq2 devtools DEXSeq dichromat digest DirichletMultinomial DNAcopy
 doMC doParallel doRNG doSNOW dostats dplyr dtw dynamicTreeCut e1071 earth edgeR elasticnet energy entropy evaluate
 evd exactRankTests exomeCopy ExomeDepth 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 ggm ggplot2 ggpubr 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 HTqPCR httpuv httr hwriter igraph illuminaio impute infotheo
 inline intervals ipred IRanges IRkernel iterators itertools jsonlite kernlab KernSmooth knitr ks kSamples labeling
 Lahman lambda.r lars lattice latticeExtra lava lazyeval LEAP leaps LearnBayes lhs limma linprog lintr lme4 lmtest
 locfit longitudinal lpSolve lpSolveAPI 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 modeltools multcomp multicool multiHiCcompare multtest munsell
 muscle mvtnorm nleqslv nlme nloptr NMF nnet nor1mix numDeriv OmicCircos optparse org.Hs.eg.db OrganismDbi pamr
 party partykit pbapply pbkrtest pcalg pcaMethods penalizedSVM peperr pheatmap 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 randomForest randomGLM rbenchmark RBGL rbokeh RCircos
 RcisTarget RColorBrewer Rcpp RcppArmadillo RcppEigen RCurl registry relaimpo relimp remotes reshape reshape2
 reticulate rex rFerns rgl Rgraphviz rhandsontable rhdf5 RhpcBLASctl Rhtslib rJava rjson RJSONIO rlecuyer RMariaDB
 rmarkdown rmeta 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 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
 testthat TFBSTools TFMPvalue tgp TH.data tidyr tidyverse topGO tradeSeq 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
 nghiavtr/BPSC weililab/scMAGeCK
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, Ncpus = $NPROC, 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

# 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

# The major work is done, the remains are rather nice gifts. (maybe remove them in the next cycle)

# 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

# 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