Bring scenario_demo_2020 from r2dii.scenario

[maurolepore]

See https://github.com/2DegreesInvesting/r2dii.scenario/blob/master/data-raw/scenario_demo.R

[jdhoffa]

make sure that it's scenario_demo_2020 that comes out! it must be prepared by the year

[maurolepore]

Add scenario_demo_2020

• Add "data-raw/scenario_demo_2020.csv"
• Add "data-raw/data_dictionary/scenario_demo_2020.csv"
• In "data-raw/scenario_demo_2020.R", add something like:

  # Source: @<contributor> <URL to issue or pull request>
  scenario_demo_2020 <- readr::read_csv(here::here("data-raw", "scenario_demo_2020.csv"))
  use_data(scenario_demo_2020, overwrite = TRUE)
  

• usethis::use_r("scenario_demo_2020"). Document scenario_demo_2020
• usethis::use_test("scenario_demo_2020"). Add regression tests of scenario_demo_2020
• source("data-raw/data_dictionary.R"); devtools::load_all()
• usethis::use_test("data_dictionary"). Test data_dictionary includes scenario_demo_2020

--

@jdhoffa:

1. Can you add the data dictionary?
2. Do we need raw_scenario_demo?
3. Can you describe scenario_demo_2020 in its helpfile?

Also,

a. The code we use to create scenario_demo uses r2dii.scenario, which is still in early development and might change in ways that compromise the reproducibility of scenario_demo. I'm not sure what's the best approach. Maybe we need to bring the relevant code to r2dii.data/data-raw until r2dii.scenario is more stable? Or we may tag a good commit of r2dii.scenario and create a github release.

[jdhoffa]

1. Done
2. We do not, I want to work on this file a bit more before including in r2dii.data
3. Almost complete.

TODO:

• Add @format to scenario_demo_2020.R (I wasn't sure how you systematically generate this from the data_dictionary entry, could you show me?)

a. r2dii.scenario is unstable and r2dii.scenario::raw_scenario_demo is unstable, however scenario_demo_2020 is stable (in my opinion). Any changes to how we generate this file shouldn't change the format of the file, this is why i felt comfortable including it in r2dii.data. We could try to remove the dependency on r2dii.scenario::add_market_share_columns() altogether by including only scenario_demo_2020.csv itself in data-raw, what do you think?

[maurolepore]

We could try to remove the dependency on r2dii.scenario::add_market_share_columns() altogether by including only scenario_demo_2020.csv itself in data-raw, what do you think?

I'm not worried about the dependency because we call from data-raw/, which isn't checked by R CMD check. I'm worried about not being able to reproduce scenario_demo_2020.csv later on, or to reproduce but with a lot of work, like digging in the git repo of r2dii.scenario for a commit that can indeed reproduce.

To be safe and clean I suggest this:

@jdhoffa:

• Find a good commit where the r2dii.scenario with the relevant code passes R CMD check cleanly and share that commit with me.
• Increment the dev version number, tag the commit, and create a github release. I can do this but I suspect you'll enjoy playing with use_github_release() or

git tag -a 0.0.900x -m "Tag good commit for scenario_demo_2020"
git push upstream 0.0.900x

Then https://github.com/2DegreesInvesting/r2dii.scenario/releases

me:

• Copy the relevant code to r2dii.data and commit. This ensures that whatever happens with
  r2dii.scenario (even if we kill the repo) the code will be easy to recover.
• Remove the relevant code and point to the code in r2dii.scenario as we do now, except that
  we can link to the specific github release (i.e. git tag).
• Sleep well.

[jdhoffa]

Release 0.0.9001 of r2dii.scenario now stable-ish (passes R CMD check) boo-yah.