Consider splitting up build steps so they can be run independantly

[alistairjcbrown]

1. Pull data from sites and transform
2. Hydrate data by finding matches on moviedb
3. Generate calendar files
4. Generate combined data blob and site

Currently 1-3 are all in the "Generate Calendar' workflow

Problem -- hydration and crawling are linked meaning

• An issue hydrating will fail the run, even though the data is good
• Changing how we match for hydrated data requires crawling the sites again

Instead could we split these workflows so that

• 1 -- create site data only release
  • → 2 -- create hydrated site data only release
    • → 3 -- generate calendar files from hydrated release
    • → 4 -- generate combined data blob and site from hydrated release

If we change the matching, we just re-run from workflow 2 (which will auto re-run 3 and 4)
If we change the calendar, we just re-run from workflow 3
If we change the site, we just re-run from workflow 4

[alistairjcbrown]

Questions:

• Do we need separate repos to deal with separate data?
  • Would this be an opportunity for a new Github organisation to house this all
  • Are there issues with code sharing if we do this?

[alistairjcbrown]

Possible plan

1. Figure out a good name
2. Create a new Gtihub org
3. Repos for:
   • Grabbing cinema data and publishing release
     • We can even split up the runs so that each group of cinemas runs on a separate runner
   • Hydrating cinema data with moviedb information and publishing release
   • Generating calendar feed and publishing release
   • Generating website and deploying
   • Generate calendar website and deploying