Notes on large.Rmd

[maelle]

• Maybe add a few sentences explaining that with large data you want (actually I think these could be the three sections of the vignette)

  • input data out of your RAM (so you're probably using a file format that's good for large data, and you're in luck there are functions for inputting this to duckplyr)
  • efficient computation (yay DuckDB, and here you don't even need to learn syntax beyond dplyr)
  • output also out of your RAM unless small (funnel, compute to file)

• Drawback of large data + duckplyr is that the limits of duckplyr won't be made up for by fallbacks since fallbacks to dplyr necessitate putting data into RAM.

• In the latter case, if too many fallbacks needed, do we recommend using dbplyr?

• The paragraph on dbplyr should be in the README.

• Some functions described in large.Rmd, in particular duckdb_tibble(), don't seem relevant for large data?