ivancea / autostocklist Goto Github PK

Reorganize directories:

• database: Docker and SQL
• server: Rust
• web: React

Easier deployment and testing.
Compile and copy files in dockerfile. Use Actix to serve the statics.

Instead of calculating the stock from the loss/resupply tables:

• Save the stock in the item table
• For a loss/resupply:
  • Open a transaction
  • Modify stock (a CHECK constraint would keep it positive)
  • Insert/Update the loss/resupply table (a CHECK constraint would kep them positive too)
  • If any of those fails, ROLLBACK and report
  • If both work, COMMIT

Options:

• React SPA
• Rust WASM:
  • Seed: Macro syntax (https://github.com/seed-rs/seed)
  • Yew: HTML/JSX like syntax (https://github.com/yewstack/yew)

Oauth with Google for MVP?

Data per day: Consumed stock, current stock

Controller methods must be moved to other modules and reorganized.

Use configure and scope todelegate configuration todifferent modules and add custom paths (https://actix.rs/docs/application/).

Edit

web::Scope avoided because of it swallowing requests its services can't process. See: actix/actix-web#2392

Final

User -> Stock list -> Item -> Stock movement

User {
   /* Auth, basic information */
}

Stock list {
    name.
    default?
    /* Future: buying and analytics configurations */
}

Item {
    name,
    minStock,
    maxStock,
    actualStock
    /* Future: buying and analytics configurations, price (for spending analysis) */
}

Stock loss{
    date,
    quantity (representing a stock reduction)
}

Stock resupply{
    date,
    quantity (representing a stock increase)
}

Stock total (VIEW) {
    itemId,
    stock
}

MVP

• No users
• No multiple lists

For a PK of (user, item, date), it'll generate around 365 light entries per user/item, in the worse case of all items consumed every day.
For an average of 50 items per user (big shopping list), it'll go around 15.000 entries per year per user.

15.000 entries per user, knowing that a RDBMS is required for the users, items and configurations management, using ScyllaDB may be an overkill, and a development obstacle as it has a slower startup and worse tools to interact with.