Roadmap

[yongrenjie]

Roadmap

Major features already in cygnet

• Directory navigation with cd
  • "Normal" behaviour with directories
  • cd - to return to previous directory
  • Error handling for directories which don't exist
• A command parser for the programme itself
• The quit command
• The help command
• Switch to (strict?) Data.Text
• A representation of the article datatype
• The read and write commands: interaction with YAML
  • Parsing cygnet.yaml (or abbotsbury.yaml) into a list of articles
  • Making sure that this is done every time the directory is changed
  • Saving upon exiting the directory / the programme. I think cygnet's backup system has generally been unnecessary.
• The list command: pretty-printing articles
• The open command: file system management
• The sort command
• The cite command
• The add command
  • Fetching data from Crossref
  • Implementing asynchronous HTTP for multiple requests
• The delete command
• The edit command
  • Tempfile creation
  • Calling vim (or $EDITOR) from inside the program
• The fetch command
• The search command

Less important features

• The import command. Quite rare, my usual workflow is to add the DOI, then fetch and open, rather than to download a PDF from the web. Still, this is helpful with (for example) emailed papers.
• The update command. I don't remember the last time I manually did this.
• Manual PDF management
  • addpdf
  • deletepdf

Other stuff

• abbot cite subcommand (right now cygnet uses a cygnet-cite entry point which is clunky)
• Other datatypes, particularly...
  • Books
  • Theses
• Pipes, allowing search results to be piped into cite (for example). We should plan this ahead of time, and have commands like search or list return two things: (1) a list of references, (2) text that can be printed to stdout. If it is part of a pipe, then we can pass (1) to the next function, and if not, then we can just print the text.
• Better megaparsec usage, including better command identification and better error reporting.
• A new command. This would let us enter metadata manually, and is important for books which have not very good metadata in Crossref.
• A command for "quick" tagging (technically can be done using edit, but we would usually like to tag something without going into vim).
• A search_tag command (abbreviated to s_tag, st, or similar), which only searches in tags.
• Let search work with acronyms. We can do this crudely by "abbreviating" the title (taking the first letter of each word) and adding it to the haystack.
• Don't destroy user input from new if the parsing fails :-(
• Modify book to use subtitles as well