Add motivation to introduction

[gcapes]

The only objectives form the index page which aren't already covered in episodes or the intermediate tutorial are these:

convert a shell script into a CWL workflow (lesson-level objective)

• explain the difference between control flow and data flow
• identify the inputs and outputs, tasks, and links in a script
• (recognize and remove details and configuration in a script that are specific to particular infrastructure)
• (remove loops for moment, these will be covered in intermediate lesson with scatter)

I've tidied up the index page in #72, but I think it would be very useful to cover at least the first two objectives above in an episode. Indeed the file name of episode 2 02-shell_to_cwl.md suggests this was once the plan. Currently the introduction hints at why you might choose to use CWL over shell scripts. But I always like it when the motivation is clear, like in the make and git lessons. e.g. without this tool, this is what you have to do. That's not ideal because x, y, z - this tool fixes those problems. This currently sticks out as an omission

http://swcarpentry.github.io/make-novice/01-intro/index.html http://gcapes.github.io/git-course/01-introduction/index.html

[mr-c]

Thanks @gcapes ; you might find motivation for "why use workflows" in https://arxiv.org/abs/2105.07028

[melibleq]

Preview of the new website: https://deploy-preview-75--cwl-website.netlify.app/ - the "Features" content might be especially relevant here

[gcapes]

I didn't find either resource gave enough detail for me to be able to answer this, but a recent discussion I had was quite illuminating. Here's some things which weren't clear to me

• [ ] setup says that we're downloading tools to use

  For this tutorial some existing tools are needed to build the workflow. These existing tools will be imported via GitHub

  However, these are only the tool descriptors, rather than the tools themselves. This confused me because I thought that the point of docker is to contain the whole environment and if you need to download a tool this defeats the purpose. I think my understanding of docker is about right, but I was thrown off by not understanding (or really examining) what I was downloading.

• [ ] One of the big selling points for CWL is that it makes using docker easier. So while you don't have to use docker to use CWL, docker gives reproducibility benefits, and CWL provides an easier way to access this. It would be informative to show the equivalent docker command for one of the CWL files.
• [ ] The docker images which are pulled at run time are define in the tools rather than the CWL scripts
• [ ] Addressing the third learning objective would also be great in providing motivation. Then you could motivate a move from scripts to docker, then docker to CWL.

[douglowe]

I've added an example docker run to the start of our first lesson, to give students some experience of running containerised tools directly on the command line. See this PR: https://github.com/carpentries-incubator/cwl-novice-tutorial/pull/78

I think this might help with some of our motivation points above.

[mr-c]

Why workflows vs bash/Make?

• Switching infrastructure (local vs. HPC; or switching HPC schedulers, or moving to cloud)
• Easier to use software containers

use the above with examples, also mention these below

• Sharing is easier
• Scaling is easier
• Better reporting (management of complexity), and ‘for free’
• Job caching (re-use of previous runs)
• Simpler to reuse components (sharing sub-workflows, tool descriptions)