Add documentation

script/README.md

@@ -0,0 +1,52 @@
+# Type-stable wrappers via tibblify
+
+Goal: A `cuf_*()` function for each `cu_()` function that returns a tibble with a known structure.
+
+## Working examples
+
+- [x] `cu_get_folders()`
+- [x] `cu_get_lists()`
+- [x] `cu_get_tasks()`
+- [x] `cu_get_time_entries_within_date_range()`
+- [x] `cu_get_singular_time_entry()`
+- [x] `cu_get_list_members()`
+
+## Building blocks
+
+- `R/df-*.R`: started as copies of `R/api-*.R`, need adaptation
+- `script/tibblify-??-*.R`: scripts that access the API to generate specs from data
+- `R/spec-*.R`: generated from `script/tibblify-??-*.R`
+-  `R/tibblify.R`: tooling
+
+## How to extend
+
+- Pick an entity type
+- Make a copy of an existing `script/tibblify-??-*.R`
+- Adapt to collect data matching this entity type
+- Run the script
+- Adapt the `cuf_*()` functions following existing examples
+- Test, perhaps as part of the same script (after another `pkgload::load_all()`)
+
+## Test instance
+
+For testing, I created a throwaway ClickUp workspace that can be populatd with the `create.R` script.
+The data in a live workspace is richer, but modification and deletion can only reliably be tested on a toy workspace.
+The "team ID" is hard-coded in `create.R` to ensure that the code doesn't accidentally wipe the live workspace.
+
+## Technical debt
+
+- Fix warnings in `devtools::document()` -- delete first four lines in most `df-*.R` files
+- Enhance `create.R` script
+    - Rerun `tibblify-all.R`, ensure that the generated specs are not worse than the existing specs
+    - Eliminate all instances of `lcol_guess()` in the specs by creating a richer example workspace in `create.R`
+- Add tests against toy workspace
+
+## Problems
+
+### Deletion
+
+`cu_delete_*()` often returns an empty list. It feels more consistent to keep the wrappers here, even if they return an empty tibble.
+
+### Visible vs. invisible return
+
+Should `POST` and `PUT` methods return invisibly?