advisor.documentation¶
Mock, alpha (0.1.0a1): a placeholder for a persona of the agent roster (docs/specs/agents.md), and the
stub drafter of a project's README and data dictionary (#200). It gives no real advice and is
likely to change shape.
The Documentation Editor drafts a README and a data dictionary for a dataset (Blueprint R5). It serves two steps of the process flow:
| Step | What the mock does |
|---|---|
setup.documentation |
Drafts a README or a data dictionary from the project's file inventory. |
preparation.documentation |
The same, for documentation drafted or reviewed while preparing the deposit. |
Input and output¶
It accepts the core DraftRequest and returns the core Document, so its drafts become project
documents (Projects) and the real editor can replace it
without its caller changing. A request names the document_kind:
readme: a Markdown README with the sections ofreadme.yaml. Each section shows, in italics, what the researcher should write in it. The Files section lists the inventory's files with their formats and sizes. Variables lists each table's fields with their types. Access and licence lists the licences the chosen repository offers, if its requirements name any.data_dictionary: a CSV with one row per field of each table in the inventory. The columns arefile,field,type,format,description,unitandmissing_values. The file, field, type and format are copied from the inventory, and so is a description where the inventory has one. The unit and missing value codes are always left empty.
The README's sections are the common core of the research data README templates compared in
200: Cornell's Data README, Dryad's and the Borealis working group's minimum README.¶
It abstains with insufficient_input when the request has no inventory, or when a data
dictionary is asked for and no file in the inventory has fields. It abstains with
outside_agent_scope for any other kind. When the request gives a current_version, the mock
drafts afresh, names that version in revises and keeps its title.
Its grounding mode is none: it cites only its input. The core Document has no limitations
slot, so the outcome's statement says what the mock does not do (docs/specs/agents.md).
What it does not do¶
- It writes no prose of its own and invents nothing. The title, description, methods, licence and citation are left to the researcher.
- It does not open the files, so it says nothing about their contents beyond what the inventory records.
- It infers no variable definitions, units or missing value codes.
- It does not read an earlier version, or the reason a person gave for a decision on it.
- It follows no data dictionary standard, such as CDIF's variable descriptions (#220).
The real editor, drafting with a model from the files and the project record, is TODO(#358).