Decide how faithful we want to be to the API

[richierocks]

If we want to be strict about having exact 1 to 1 mappings with the API, we should have

1. list_projects() that calls GET /projects, and
2. get_project() that calls GET /projects/project-id

The latter isn't that useful for data science, since I suspect you'll actually want to retrieve lots of projects at once to compare. I suggest instead having a higher level get_projects() that can return multiple projects at once. I think the pattern is:

#' @export
list_projects <- function(workspace, team, archived = FALSE) {
  # some input checking
  results <- call_asana_api("projects/", workspace = workspace, team = team, archived = archived)
  # some flattening ...
  flat_results
}

# internal only
get_project <- function(project_id) {
  # checks
  results <- call_asana_api("projects/", project_id)
  # flattening
  a_single_row_data_frame
}

#' @importFrom purrr map_df
#' @export
get_projects <- function(project_ids) {
  map_df(project_ids, get_project)
}

[chriscardillo]

I agree!

I have a question for this specific example - there are three options for querying projects:

1. Query all projects via /projects endpoint
2. Query workspace-specific projects via /workspaces/workspace-id/projects
3. Query team-specific projects via /teams/team-id/projects

So would we then have 3 functions, specifically:

1. get_all_projects() for endpoint /projects
2. get_workspace_projects() for endpoint /workspaces/workspace-id/projects
3. get_team_projects() for endpoint /teams/team-id/projects

Where 2 & 3 would take some workspace_id and team_id param accordingly, as well as the archived param?