10 ppl attended

Organisational matters

• the next meeting will take place on the 10th January

• Francesco Carrer will give a short intro on the tools for spatial statistics he’s been developing

Conferences, meetings and sessions

• CAA2022: The conference will probably take place in July (announcement via twitter ), but no exact dates have been released yet

SIG activities

• Teaching material list: Sophie Schmidt has shared a google document in the respective slack channel. Please invest some minutes of brainstorming to add the relevant materials you know.

Discussion

• interesting blogs
  • LaTeX Ninja
  • Alison Hill on using blogdown
  • Ben Marwick on using Atom

Show and tell

• Show and tell focused on pipeline tools

• Compiling software projects is similar to running data analysis scripts, one needs to run a number of scripts / functions in a certain order

• Therefore there are pipeline tools for any kind of scripting language “out there”

• Several pipeline tools can be found here: https://github.com/pditommaso/awesome-pipeline

• Ben Marwick introduced the R packages {targets} and {renv}

  • {targets}
    • if you use Rmarkdown and don’t want all analysis steps to run every time you knit and dislike the caching knitr offers
    • target “flips” the process
    • the analysis steps are outsourced to functions in separate scripts
    • _targets.R is the “recipe book” similar to a make-file that defines the run order of the scripts
    • at the end the document is knitted
    • this way: unchanged scripts are skipped (and there is a load function for data, which will “alert” the later scripts if the data has been changed)
    • in the RMD you only reference to the images you create with the scripts
    • especially well suited for tasks that take long to compute
    • active maintainer
    • corresponding article: https://doi.org/10.21105/joss.02959
    • tutorial: https://wlandau.github.io/targets-tutorial/#1
  • {renv}
    • “better packrat”
    • saves your package versions in a renv.lock file
    • the Rprofile file sources renv action, which compares the versions of your session / project which those given in the renv.lock
    • alerts you to changes and uses the package versions given in renv.lock, but doesn’t overwrite any of your packages (if you close the project, everything returns to default)
    • three functions needed: renv::init(), renv::snapshot() and renv::restore()
    • easier than the rrtools workflow using the DESCRIPTION file

• Clemens showed the haskell based Shake and described the same steps as detailed in his blog post

  • shake only needs stack
  • within a shakefile the rules in which order to run which scripts and what are input and output data are defined
  • Clemens created some operators to reduce the complexity of the rules syntax in a shakefile

Next meeting: Monday, January 10, 2022

• everyone: Happy holidays and merry Christmas to everyone celebrating!
• Have a great start into the new year, looking forward to seeing you again!