Skip to content

TODO.md

Latest commit

 

History

History
66 lines (42 loc) · 3.1 KB

TODO.md

File metadata and controls

66 lines (42 loc) · 3.1 KB
Raw

TODO

In general

  • Better structure of tests with less specialized cases?
  • Update the couch-db action a bit? See problems with _changes etc., and different configurations and contents.
  • rename uidlist to index everywhere because that is what it is. (Not sure I am done with that.)
  • Rewrite of simplify: Okay. But now think about the missing tools / utilities and vignettes to advertise them?
  • Reduced functions, make it easier to maintain somehow. (HAHAHAHAHAHAHA)
  • Clean up old code. (Ha.. )

Maybes

  • Some possibility to only update resources that have been changed since the last synch since loading everything is excessive for large projects.
  • Getting CRAN-ready: Better structured tests that do not depend on DB-connections.
  • Reduced messaging behaviour.

March Madness Cleanup TODOs according to Claude

💀 Dead Code (Not Called Anywhere in the Pipeline)

convert_to_onehot() — Exported, has a test, has docs. But simplify_idaifield() never calls it. It was part of the old spread_fields behaviour, which was deprecated. It's just sitting there. It wants attention.

fix_dating() and bce_ce() — They're exported, tested, documented — but completely orphaned from the actual simplification pipeline. The dating fields just fall through to the generic unlist() at the bottom. It wants attention.

idf_sepdim() — You called this "VERY VERY BAD" yourself. It's exported, never called internally. Dead.

🔨 Broken (Documented as Such)

:)

🤔 Questionable Exports

gather_languages() — Exported, has tests, but the language selection in simplify_single_resource() is entirely commented out. So it's a public function with no path through the package to reach it. Power users could call it directly, but is that the intent? Big todo to fix this!

find_named_list() — Exported, has a test. Never called internally. Appears to be a utility for users to dig into the config list. Fine if intentional — but is it? -> Is used now to find the resource list, I should keep it but make the doc more explicit on how one can use it.

select_by() — A deprecated wrapper for idf_select_by(). Fine for backward compat, but it lives in select_idaifield.R instead of deprecated.R where it belongs.

🧹 Minor Mess

proj_idf_client() is still chatty. Still prints "Connected to project 'X' containing N docs." on every internal call. Every query, every get_field_index(), everything goes through this and yells at the user.

idaifield_as_matrix() example still uses select_by() (deprecated). Should be idf_select_by().

Aesthetics

Find a better naming scheme for all functions and rename / deprecate accordingly.

Summary Table

Item Status Action
convert_to_onehot() 🧹 Stale Document as standalone tool and use again
fix_dating() + bce_ce() 🧹 Stale rethink into a better tool and then advertise
idf_sepdim() 💀 Dead Rewrite or remove
proj_idf_client() chattiness 🧹 Annoying Move message out
select_by() example 🧹 Stale Update to idf_select_by()