RSE 2018 Shiny Workshop

This repo contains the website, slides and other material for the RSE 2018 Shiny Workshop.

The website for the workshop is at https://uomresearchit.github.io/RSE18-shiny-workshop/

The code required for the running example is in a separate repository at: https://github.com/UoMResearchIT/RSE18-shiny-workshop-materials . This is cloned to each user's (virtual) machine.

Each step of the exercises is a separate tagged commit in this repository. These commits are linked to the relevant github commit page on the workshop slideshow and on the website.

Build/modification process

To edit the webpage or slide show, edit the relavant .Rmd file. Do not edit the .md files - these are automatically generated. make will build the site and slideshow as required. The build process for the .Rmd files is:

1. Resolve links of the form [git:tagname]() to the appropriate page on github. This step will fail by design if there are any unresolved links (i.e. there isn't a commit with tag "tagname" in the coursematerial repository)

2. The .Rmd files will be "knitted" to .md files ready for Hugo. R code snippets within them will be executed and the output included in the .md file.

hugo serve will let you locally preview the changes. make cleancontent will delete all the generated .md files. It's safer to stop the local server before running this and regenerating the content if needed; otherwise the left hand menu isn't generated reliably.

Push the repo to github to deploy.

Modifying the exercises

The aim is to keep a clean commit history in the workshop-materials repoistory, with each commit corresponding to a solution to an exercise, or a check point in the lesson. If you find a bug in the code, or need to modify something, we rewrite the history to make it look like the bug was never there. The basic procedure is:

1. Checkout the commit where the issue is introduced
2. Create a new branch git checkout -b bugfix
3. Fix the bug
4. git rebase bugfix master
5. git branch -d bugfix (At this point you should have a linear sequence of commits, including the extra commit where the bug was fixed, plus a (buggy) branch with the tagged commits)
6. git rebase -i --root to interatively rebase the (good) branch. Squash the bugfix commit with its parent
7. Re-tag the commits on the new branch (git tag -d tagname, git tag tagname commitid). This is manual at the moment (TODO automate this - should be possible to match on commit messages, as these will be identical on the buggy and good branch)
8. Check you have a linear commit history with (all) the commits tagged
9. Regenerate the website / slides to use the new commitids: make cleancontent, make
10. Force push the coursematerial website (at this point the old commits will no longer be visible in the log since nothing points to them)
11. Force push the (new) tags to the coursematerial repo: git push --tags --force
12. If you've ended up with tags you're no longer using on the remote repository, delete these on the remote with git tag -d oldtag, git push origin :refs/tags/oldtag (see https://nathanhoad.net/how-to-delete-a-remote-git-tag)
13. Push the main repository.

Editing the gapminder data

The gapminder data used in the course is different from that included in the gapminder R package. The data in this course has values for every year rather than the 5 year intervals uses in the gapminder package. The createData.R script pulls together the source data (downloaded from the gapminder website), deals with standardising country names, etc.

The data included also contains total fertility rate and subregion - I ended up not using these in the course, but I've left them in in case we use them when we extend the workshop to a Research IT training course.

Old notes on Docker image

Early work was done in a Docker image - this is no longer needed, as it was more space-efficient to install RStudio etc directly in the virtual machine supplied by the organisers. Build instructions for this, and a test script are in vminstall

The Docker image can be built with make dockerimage The Docker image can be started/stoped with run|stoprstudio The slide deck will be compiled and opened with make present

If running the Docker image, connect to localhost:8787, username/password: rstudio/rstudio

We also map port 3838 so we can view Shiny apps in the external browser.

Note the RunApp button doesn't seem to work properly in Firefox - the app doesn't initialise properly. Is OK in Chromium.