dependencies.Rmd

# Dependency Management {#dependencies}

*How do I make sure that all my software and configurations needed for a projects are portable?*


- [renv](https://rstudio.github.io/renv/articles/renv.html) captures package versions and sources as well as R version. `renv` also records python versions and virtual environments for projects that used both languages. 
    - The [capsule](https://github.com/MilesMcBain/capsule) package can be useful when you want version control/depedency management but not the full weight of `renv`. Sometimes, `renv` can be an impediment for smaller and highly collaborative projects, that is where capsule can be useful. Ultimately, capsule will allow to setup `renv` for a project when it gets to a stable state.  

-   [Docker](https://www.docker.com/) allows you to create virtual machines, meaning you can capture all system dependencies in addition to the dependencies captured in `renv`. It works for essentially any programming language and is extremely flexible. 
    - [UseR 2022 Docker workshop](https://github.com/rsangole/user2022-r-for-docker/tree/main)
    -   [A lesson in user Docker for an R
project](http://ropenscilabs.github.io/r-docker-tutorial/)
-   Makefiles can automate a complex, multipart project. Here's [a lesson on them from Software Carpentry](http://swcarpentry.github.io/make-novice/)

-   R packages can be a useful project output. We have some in-house R packages to provide access to internal data and generate reports, and may be developing more for external audiences. Hadley Wickham's [R Packages Book](http://r-pkgs.had.co.nz/) provides guidance for these, and we expect our packages to be up to [rOpenSci standards](https://github.com/ropensci/onboarding).


##  GCC and mac silicon

-   R package installation can be especially tricky on Mac computers with [Apple Silicon](https://support.apple.com/en-us/HT211814). Below are some steps to get 
    package install set ups.
    
1. Install the R compilation tool chain https://mac.thecoatlessprofessor.com/macrtools/. If any issues with FORTRAN, follow instructions for downloading FORTRAN compiler here for Apple silicon Macs: <https://mac.r-project.org/tools/>

2. Install gcc `brew install gcc`. This requires having brew set up. 

3. Make sure homebrew is in terminal path. Add it by `export PATH=$PATH:/opt/homebrew/bin`.

4. Create `~/.R/Makevars` and insert the following. This will need to be updated depending on gcc version. 

```
CXX14FLAGS += -O3 -arch arm64 -ftemplate-depth-256
FC = /opt/homebrew/Cellar/gcc/13.2.0/bin/gfortran
F77 = /opt/homebrew/Cellar/gcc/13.2.0/bin/gfortran
FLIBS = -L/opt/homebrew/Cellar/gcc/13.2.0/lib/gcc/13
```

5. RStudio may have a bug in accessing the correct terminal path. The workaround is to copy your terminal path (`echo $PATH`) into `.Renviron` per instructions here: <https://community.rstudio.com/t/how-to-get-rstudio-ide-to-use-the-correct-terminal-path-in-mac-os-x/131528/6>

6. Some packages may need additional help for R to find the correct paths. For `units`:
`remotes::install_version(package = "units", 
                         version = "0.8-5", 
                         configure.args = "--with-udunits2-lib=/opt/homebrew/Cellar/udunits/2.2.28/lib --with-udunits2-include=/opt/homebrew/Cellar/udunits/2.2.28/include")`


##
Historically some projects used [Packrat](https://rstudio.github.io/packrat/) or [checkpoint](https://CRAN.R-project.org/package=checkpoint) to fix R package versions. These systems have been superceded by `renv`.