TODO.md

TODO

Logic

stateful logic

• make Impossible States Impossible 🎉 🎉 🎉
  • managed certainty of type Status and type Output
    • Status: can only be Loading, Loaded something, and Errored
    • Output screen: can only show one line at the time, depending on user input
• avoid boolean identity crisis 🥳 🥳 🥳
  • see type Sound

keyboard

• display keyboard onscreen
• onclick events to output
• onclick events to sound
• use an package to get user typing
• get keycode from typing into onscreen keyboard
• filter out KeyPressed to
  • if alt/meta/shift/ctrl/repeat is pressed do nothing
  • allow alphabet characters
  • allow commands keyboard shortcuts
    • FIX: Sumbit It keyboard shortcut doesn't use the right word
  • send keycode to screen (see "user feedback" section)
  • send keycode to speak (see "user feedback" section)

text to speech 🤖🤖🤖

• SpeechSynthesis Web API with a JavaScript Port
• let user disable voice
  • completely
  • for some features (no. API only pauses/resumes global state)
    • just keys
    • just feedback
• no need for Sub (for now?)

refactor

• refactor update and use helper functions
  • reuse helper functions to streamline code
  • declutter view's onClick events and use update for Msg
  • not all Msg need to be parametrized

output screen

• use a Maybe String for model.click to
  • show default message or GuessWord
  • implement a conditional to only show GuessWord if valid
  • awlays show clicked Nothing if GuessWord is empty
• switch to using type Output with a case ... of
  • much clearer and less convoluted code
  • it can either be Init, Holder, Word, or Result upon sumbit
  • avoid double use of Maybe Strings that depended on one another
• has user started entering a word?
  • no: display default messasge telling the user what to do
  • yes: display typed text
• erase letter: if guess word is emtpy display default message
• retry/reset: display default message
• submit: result text to take over as the only text on screen

UI/UX

• look like Speak & Spell (make it look & feel like a toy)
  • keys (alphabet, commands) to look different
• user feedback
  • feedback for keypresses
    • written feedback
    • auditory feedback with SpeechSynthesis
    • visual feedback (change color)
    • is key stuck?
  • feedback upon checking the spelling
    • written feedback
    • auditory feedback
• screen
  • retro-looking/low-fi/8bit
  • fixed "field" for the user output
  • Elm Instruments
  • add logo besides "brand"
• create a custom loading page
  • use elm logo :)
  • spell LOADING with Speak & Spell styled keys
  • animate the LOADING words on screen
• create custom error page
  • implement function to handle http.error
  • and return custom (parsed) error messages
• ARIA accessibility
• SEO optimization
• make elm-ui design responsive
  • elm ui seems abandonware
  • making it responsive is a bit PITA
  • leaving as is and will do proper CSS (elm-css?) reponsive design later

Improvements

• Elm optmization:
  • remove debug.string + refactor those fn
  • use lazy
  • assets size
  • use local fonts
• use parceljs

Maybe

• make it so that KeyPressed triggers visual feedback on keyboard keys
  • should I use a Maybe KeyboardEvent?
  • refactor Msg and split into KeyPressed + KeyClicked types?
    • would this be too convoluted? Elm likes clarity over complexity

Provided that a polyglot API (or separate different languages APIs) exist, these are a number of items I'd like add to the app:

• let user choose:
  • a different dictionary language
  • the voice language, params, and gender
• internationalization depending on chosen dict language
  • UI
  • commands

Deploy

Once complete, deploy a static website:

• deploy to GitHub Pages
  • gh pages sucks!!!
  • c'mon microsoft... learn from gitlab.
• vercel app :)