I often use LocalStorage (or SessionStorage) to store structured data on the client for demos and real projects. While this worked fine, I’ve always had a nagging feeling that there would be a better way than `JSON.parse` and `JSON.stringify` large arrays on every change. Let’s have a look at why IndexedDB might be a better choice for you than the Storage API.

  The LocalStorage and SessionStorage are herded under the Web Storage API. I will compare the Web Storage API to the IndexedDB API in this article. While you’ll see that IndexedDB is far richer in features, I will try to stick to the basics to meet my needs in storing structured data persisted in the user’s browser.

  ## Oranges and Apples

  In essence, with the Storage API, you’re limited to a get and set method where you can pass a key and a (string) value:

  ```jsx
  window.localStorage.setItem(
    'tasks',
    JSON.stringify([
      {
        id: 1,
        title: 'Learn JavaScript',
        completed: true,
      },
      {
        id: 2,
        title: 'Learn Vue',
        completed: false,
      },
      {
        id: 3,
        title: 'Build something awesome',
        completed: false,
      },
    ])
  )

  const tasks = JSON.parse(window.localStorage.getItem('tasks'))
  ```

  This is great for simple key-value storing, but not so much for more structured data. A quick comparison:

  | Feature         | Web Storage                        | IndexedDB                                    |
  | --------------- | ---------------------------------- | -------------------------------------------- |
  | Data Structure  | Key-value pairs                    | Object store, key-value pairs                |
  | Storage Limit   | Typically around 5MB per origin    | Larger storage capacity (varies per browser) |
  | Querying        | Limited querying capabilities      | Advanced querying with indexes               |
  | Transactional   | No                                 | Yes                                          |
  | Indexing        | No                                 | Support for indexes                          |
  | Data Types      | Limited to strings                 | Supports complex data structures             |
  | Browser Support | Widely supported across browsers   | Widely supported across browsers             |
  | Data Retention  | Persistent                         | Persistent                                   |
  | Use Cases       | Storing simple data, small amounts | Storing structured data, larger amounts      |

  More simply put, the storage API is a key-value store while IndexedDB is an actual database. So why do we try to jam everything in the key-value store? Well, it’s easy! The path of least resistance is often chosen while it might make the actual code for updating and deleting items more complex.

  ## Using IndexedDB

  Like every technique I like to learn, I first try to create an actual demo so I run into real-world requirements and issues while building. This time, I went for a good old todo app that persists data in the user’s browser:

  ![Screenshot of a todo app with a form and a list of tasks](/img/articles/indexed-db-web-storage-todo-app.png)

  You can [view the demo here](https://indexed-db-demo.davebitter.com/) and [view the source code over at GitHub](https://github.com/DaveBitter/indexed-db-demo).

  As I’m going to add quite a bit of functionality for the IndexedDB database, I’ll first create a helper in a separate file structured like this:

  ```jsx
  let openRequest

  export const db = {
    // Open the database and create the tasks object store if needed
    openDB: () => {
      return new Promise((resolve, reject) => {})
    },
    // Add a task to the database
    addTask: (task) => {
      return new Promise((resolve, reject) => {})
    },
    // Get all tasks from the database
    getAllTasks: () => {
      return new Promise((resolve, reject) => {})
    },
    // Update a task in the database
    updateTask: (task) => {
      return new Promise((resolve, reject) => {})
    },
    // Delete a task from the database
    deleteTask: (id) => {
      return new Promise((resolve, reject) => {})
    },
  }
  ```

  With these methods, I should be able build all the functionality the todo app. Let’s go over them one by one.

  ### Opening the database store

  With IndexedDB you need an object store for data. You can see this as a collection in your database. I need to either retrieve an existing tasks store or create a new one if it doesn’t exist yet. You can do this like this:

  ```jsx
  let openRequest

  export const db = {
    // Open the database and create the tasks object store if needed
    openDB: () => {
      return new Promise((resolve, reject) => {
        openRequest = indexedDB.open('tasks', 1)

        openRequest.onupgradeneeded = () => {
          const db = openRequest.result
          if (!db.objectStoreNames.contains('tasks')) {
            db.createObjectStore('tasks', { keyPath: 'id', autoIncrement: true })
          }
        }

        openRequest.onsuccess = () => {
          resolve(openRequest.result)
        }

        openRequest.onerror = () => {
          reject(openRequest.error)
        }
      })
    },

    // ...
  }
  ```

  First I create a variable called `openRequest` to store the opened database to be used in other methods coming up next. Then I call `indexedDB.open('tasks', 1)` and store it in the variable. The first parameter is the name of the store, the second one is the version of the database.

  You can use `onupgradeneeded` to create object stores. There I check wether it already exists or I need to create one. When creating, you notice that I pass a `keyPath`. By specifying that the `keyPath` is the `id` property of the task, I can later easily update or delete a task by passing the `id` of the task. By passing `autoIncrement: true` IndexedDB will automatically fill this property incremented based on the highest `id`. I can call `onsuccess` and `onerror` to handle whether the object store was created successfully or not.

  Great! I now have an object store to work with! Let’s add the actual methods allowing me to perform CRUD operations.

  ### Creating a new task

  Firstly, I want to be able to add a new task to the object store. I’ve build a form to gather the data and now need a method to pass the task object to:

  ```jsx
  let openRequest

  export const db = {
    // ...

    // Add a task to the database
    addTask: (task) => {
      return new Promise((resolve, reject) => {
        const db = openRequest.result
        const transaction = db.transaction(['tasks'], 'readwrite')
        const store = transaction.objectStore('tasks')

        const addRequest = store.add(task)

        addRequest.onsuccess = () => {
          resolve(addRequest.result)
        }

        addRequest.onerror = () => {
          reject(addRequest.error)
        }
      })
    },

    // ...
  }
  ```

  First, I get the database from the `openRequest` by getting the `result` property. Next I need to create a transaction. A transaction is basically series of operations that all need to pass successfully in order for the entire transaction to be applied. If one operation fails, all the operations won’t be applied. In this article I won’t go to deep into this, but you can [read more on transactions here](https://www.dbvis.com/thetable/database-transactions-101-the-essential-guide/#:~:text=A%20database%20transaction%20is%20a,operations%20performed%20within%20a%20DBMS.). First I pass an array of object stores I want to modify. In my use case, I only need to make an update to `'tasks'`. Secondly I pass the type of operations I want to perform. As I’m retrieving and writing data I need to pass `'readwrite'`.

  Next, I get the `'tasks'`store from the transaction and I call the `add` method with the task data. Again, I have `onsuccess` and `onerror` handlers that I can use to provide feedback in my application.

  And that’s it! Note how much easier this code is compared to using the WebStorage API and having to get all the tasks, go over all of them until I find the one to update, update the task and storing the entire list again. With IndexedDB I can just call the `add` method.

  ### Reading all tasks

  The second step is to create a method to retrieve all the tasks for the object store so I can display them in the UI. Let’s create a method that does just that:

  ```jsx {13}
  let openRequest

  export const db = {
    // ...

    // Get all tasks from the database
    getAllTasks: () => {
      return new Promise((resolve, reject) => {
        const db = openRequest.result
        const transaction = db.transaction(['tasks'], 'readonly')
        const store = transaction.objectStore('tasks')

        const getAllRequest = store.getAll()

        getAllRequest.onsuccess = () => {
          resolve(getAllRequest.result)
        }

        getAllRequest.onerror = () => {
          reject(getAllRequest.error)
        }
      })
    },

    // ...
  }
  ```

  This method works much like the `addTask` method. The only real difference is that I just need `readonly` for the transaction and I call the `getAll` method to retrieve all tasks.

  ### Updating a task

  As you might have noticed in the Todo demo, I have a checkbox where I can set the status of a task to be complete or incomplete. For this, I need to be able to make an update to (the `status` property of) the task:

  ```jsx {13}
  let openRequest

  export const db = {
    // ...

    // Update a task in the database
    updateTask: (task) => {
      return new Promise((resolve, reject) => {
        const db = openRequest.result
        const transaction = db.transaction(['tasks'], 'readwrite')
        const store = transaction.objectStore('tasks')

        const putRequest = store.put(task)

        putRequest.onsuccess = () => {
          resolve(putRequest.result)
        }

        putRequest.onerror = () => {
          reject(putRequest.error)
        }
      })
    },

    // ...
  }
  ```

  The only real difference between this method and the `addTask` method is that I call the `put` method instead of `add`.

  ### Deleting a task

  Finally, I want to be able to delete a task from the database:

  ```jsx {13}
  let openRequest

  export const db = {
    // ...

    // Delete a task from the database
    deleteTask: (id) => {
      return new Promise((resolve, reject) => {
        const db = openRequest.result
        const transaction = db.transaction(['tasks'], 'readwrite')
        const store = transaction.objectStore('tasks')

        const deleteRequest = store.delete(id)

        deleteRequest.onsuccess = () => {
          resolve(deleteRequest.result)
        }

        deleteRequest.onerror = () => {
          reject(deleteRequest.error)
        }
      })
    },

    // ...
  }
  ```

  You guessed it, I just have to call `delete` with the id of the task to remove a task.

  ## Is IndexedDB worth moving over from the good ol’ simple Web Storage?

  Although they both offers persistent storage on the user’s browser, I see a clear distinction between the two. Web Storage is great for storing simple key-value pairs. There is no boiler plating needed, you just get and set items. For structured data I feel IndexedDB offers a way better API. Yes, there is more API specific code needed, but let’s have a look at actually using the Web Storage API for the todo app. You are going to need to write the logic for updating, retrieving and deleting a specific item yourself. So although the IndexedDB API requires more API-specific code, you’ll end up with roughly the same code with either of the options.

  By abstracting away the database logic into the helper I’ve shown before, you or your team still have simple methods to call in your application, but under water you are using a far more robust and structured-data specific solution to persist data on the user’s browser. Not to mention the other benefits like performance, larger storage capacity and advanced querying.

  In the end, you have to decide for yourself. The goal of this article is not to make you feel bad about (perhaps mis-) using the Web Storage API, but rather show how easy it could be with an abstraction helper to use the IndexedDB API. From now on I’ll be using the IndexedDB to persist more complex data structures in the user’s browser for my applications. Thanks for learning together!

IndexedDB > Web Storage

## Why should I use this at all?
Internationalisation is hard. There are many ways different countries handle the formatting of numbers, currencies and dates. In Dutch, we write `DD/MM/YYYY` while in English you’d write `MM/DD/YYYY`. That’s a pretty well-known one, but do you know how we format large numbers with decimals in Dutch compared to English? In Dutch we use `100.000,11` while in English you’d write `100,000.11`. These might be familiar to you but what about French? In French, you’d write `100 000,11`. There are even more differences if we look at currencies. In Dutch, you’d write `€ 100.000,11` while in German you’d write `100.000,11 €` This is where it becomes difficult to write your own formatting functions. You’d not only have to know about these differences but would have to extend your formatting functions for every language that you’re going to support.
## There is a library for that!
In the past, this would be the solution. If we look at just dates, libraries like Moment.js and date-fns have been around for a long time. Although this does the trick, it comes at a hefty price. Let’s say you just want to format some dates. Moment.js would cost you a lot of KB’s in your bundle. If you visit [Moment.JS’s website](https://momentjs.com/), you see that they actually discourage you to use their library and use the default Intl browser API instead.

> We now generally consider Moment to be a legacy project in maintenance mode. It is not dead, but it is indeed done.

This is not something to be sad about, it’s great news! Years ago Moment.js filled the gap for this need of developers that wasn’t met with a default browser API. Luckily, we do have that now.
## So what does Intl offer?
Intl offers a standard for formatting build in the browser for two main categories. Firstly we can format dates easily with `Intl.DateTimeFormat` and `Intl.RelativeTimeFormat` for regular and relative dates. Secondly, we can format numbers with `Intl.NumberFormat` for numbers and currencies. These three methods most likely satisfy the majority of your formatting needs.
### DateTimeFormat and RelativeTimeFormat
The `DateTimeFormat` method consists of three parts. The locale, formatting options and date.
```js
new Intl.DateTimeFormat('en', { day: 'numeric', month: 'long', year: '2-digit' }).format(new Date());
// outputs: "August 4, 21"

new Intl.DateTimeFormat('nl', { day: 'numeric', month: 'long', year: '2-digit' }).format(new Date());
// outputs: "4 augustus 21"

```
You can see that it will not just put the day, month and year in the correct order, but also handle the translation of the name of the month. Naturally, there is a wide range of formatting options to be used which you could check out over at the documentation.
One of the cool things in Moment.js was the ability for relative dates. You can do this to with Intl by using `Intl.RelativeTimeFormat`.
```js
new Intl.RelativeTimeFormat('en').format(-1, 'day');
// outputs: "1 day ago"

new Intl.RelativeTimeFormat('en').format(-2, 'day');
// outputs: "2 days ago"

new Intl.RelativeTimeFormat('nl').format(-1, 'day');
// outputs: "1 dag geleden"

new Intl.RelativeTimeFormat('nl').format(-2, 'day');
// outputs: "2 dagen geleden"

```
You can see that it handles the pluralisation and translation for you. This is great while working with many languages. You can check out the documentation for more examples of what you can do.
## NumberFormat
The `NumberFormat` method consist of three parts as well. The locale, formatting options and number.
```js
new Intl.NumberFormat('en', {}).format(number)
// outputs: "100,000.11"

new Intl.NumberFormat('nl', {}).format(number)
// outputs: "100.000,11"

```
Next to this, you can format the number as a currency by passing the following formatting options:
```js
new Intl.NumberFormat('en', { style: 'currency', currency: 'GBP' }).format(number)
// outputs: "£100,000.11"

new Intl.NumberFormat('en', { style: 'currency', currency: 'EUR' }).format(number)
// outputs: "€100,000.11"

new Intl.NumberFormat('nl', { style: 'currency', currency: 'EUR' }).format(number)
// outputs: "€ 100.000,11"

new Intl.NumberFormat('de', { style: 'currency', currency: 'EUR' }).format(number)
// outputs: "100.000,11 €"

```
As you can see, there are quite a few differences between locales. You can also mix and match locales and currencies to, for instance, get euro's in English number formatting.
## That's awesome!
Right!? This standardised way of formatting numbers, currencies and dates makes it a breeze to always display the correct value to users based on their locale. We don't need to load any libraries that will increase our bundle size because Intl lives right in the browser. I encourage you to try it out and use it in your (next) projects. If you'd like to see some examples you can watch my Friday tips on Intl [here](https://www.davebitter.com/friday-tips/intl) and visit the docs over at MDN. Thanks for reading!

You should use the Intl browser API

I am a Senior Front-end Consultant, Developer Advocate, and Google Developer Expert for Web. Here, you can find my resume, a collection of my articles covering a wide range of topics and get in contact.

ES6+

Full disclaimer, I came across this solution watching a 45-second video by Fireship on Youtube which you can watch [here]([https://www.youtube.com/watch?v=ITogH7lJTyE](https://www.youtube.com/watch?v=ITogH7lJTyE)). All credits to them. I'm a fan of reading these at my own pace, how impressive 45 seconds is though, so here it is in writing.
## The problem
Let's use the following example:
```jsx
const foo = async () => {
  const a = await bar();
  const b = await bar(a);
  const c = await bar(b);

  return a + b + c;
};
```
Not to bad right? Watch what happens if we introduce try-catch to handle errors:
```jsx
const foo = async () => {
  let a;
  let b;
  let c;

  try {
    a = await bar();
  } catch (error) {
    handle(error);
  }

  try {
    b = await bar(a);
  } catch (error) {
    handle(error);
  }

  try {
    c = await bar(b);
  } catch (error) {
    handle(error);
  }

  return a + b + c;
};
```
Not so pretty now. You end up with, as Fireship calls it, the try-catch tower of terror. You can of course append the catch method to the each promise like so:
```jsx
const foo = async () => {
  const a = await bar().catch(error => handle(error)); // or .catch(handle)
  const b = await bar(a).catch(error => handle(error)); // or .catch(handle)
  const c = await bar(b).catch(error => handle(error)); // or .catch(handle)

  return a + b + c;
};
```
Yes this works, but there is still a lot of repetitive code which we ideally don't want. The solution is to write a helper function that implements one try catch that replaces the others:
```jsx
const helper = async (param) => {
  try {
    const data = await bar(param);
    return [data, null];
  } catch (error) {
    return [null, error];
  }
};
```
We return an array that will contain either the data or the error. We can then use it like so:
```jsx
const foo = async () => {
  const [dataA, errorA] = await helper();
  const [dataB, errorB] = await helper(dataA);
  const [dataC, errorC] = await helper(dataB);

  return dataA + dataB + dataC;
}
```
This is looking pretty clean. Now we can also easily handle an error with an if statement:
```jsx
const foo = async () => {
  const [dataA, errorA] = await helper();

  if (errorA) {
  }

  const [dataB, errorB] = await helper(dataA);
  const [dataC, errorC] = await helper(dataB);

  return dataA + dataB + dataC;
}
```
That's it, thanks for reading!

Avoid the Async Await try-catch hell

## My problem with classes
We've all written/seen classes for adding some different states to an HTML element. A generic example with pseudo-code is the something like this:
```html
<li class="list-item">List item</li>
<li class="list-item">List item</li>
<li class="list-item active">Active List item</li>
<li class="list-item">List item</li>
<li class="list-item">List item</li>
```
```css
li.active {
  background-color: #dbdbdb;
}
```
```js
const activeItem = document.querySelector('.list-item.active');

doSomething(activeItem);
```
My first issue is with using classes as JS selectors is that I have to update my JS code as soon as I alter some classes while refactoring my HTML and CSS. In this example, that's not a big deal but in larger codebases, with more complex components and multiple team members, it can become a bit of a hassle. The Javascript is too tightly coupled to these classes which were meant to be for styling the HTML elements. I want to have my CSS and JS selectors separated.

My second issue is that there is only one value you can add to the HTML element ("active"). Implicitly, we're saying that if there is no "active" class, the styles of "inactive" should be applied. You could circumvent this by adding an "inactive" class to the HTML element, but that would mean that, technically, both classes could be applied at the same time. This can cause some styling issues you don't want. Now, this is with just active or not, but what if you have multiple states. A status component could have "success", "warning", "error", "info" and more.
## Data attributes to the rescue!
Let's refactor the above component to use data attributes instead:
```html
<li class="list-item">List item</li>
<li class="list-item">List item</li>
<li class="list-item" data-active="true">Active List item</li>
<li class="list-item">List item</li>
<li class="list-item">List item</li>
```
```css
li[data-active='true'] {
  background-color: #dbdbdb;
}
```
```js
const activeItem = document.querySelector('.list-item[data-active="true"]');

doSomething(activeItem);
```
As you can see, it's not all that different, but we get some really nice benefits. Firstly, there is a clear separation of concerns. My HTML has some classes that my CSS uses to style the list item. The data attribute shows the state of the HTML element and can be used to add the stateful styles. I am now free to change my classes in the HTML and CSS without having to worry that any of my JS breaks. Naturally, you would still have to keep an eye out for changes in the data attribute.

Secondly, there is now an explicit descriptive singular state the list item can be in. It's either active or not, true or false. This prevents some of the potential bugs I mentioned earlier with the status component example.

Thirdly, we get a nice added bonus. We can easily get all the key-value pairs out of the HTML element by calling the `dataset` property on the DOM Node:
```html
<li class="list-item" data-status="success">List item</li>
<li class="list-item" data-status="warning">List item</li>
<li class="list-item" data-status="warning" data-active="true">Active List item</li>
<li class="list-item" data-status="error">List item</li>
<li class="list-item" data-status="error">List item</li>
```
```css
li[data-active='true'] {
  background-color: #dbdbdb;
}
```
```js
const doSomething = (node) => {
  console.log(node.dataset); // { active: "true", status: "warning" }

  switch (node.dataset.status) {
    case "warning":
        doSomethingElse()
        break;
    default:
        break;
  }
};

const activeItem = document.querySelector('.list-item[data-active="true"]');

doSomething(activeItem);
```
This is great if you have further data attributes like "status" in this example. If these were classes, it would be more cumbersome to do the same.
## Final thoughts
This way of working has always been great for me to separate information needed in JS in a declarative manner and ensuring I can easily refactor components. Maybe try it out in your next side-project and let me know if this works for you as well. Thanks for reading!

Why I use data attributes over classes for declarative element states

The nullish coalescing operator is a short-circuiting operator you can use to provide an alternative for a left-hand side operand by writing `??`. You might already have used `&&` or `||` before. So how is this different?  Let's take the following examples:
```jsx
const a = '' || 'unknown' // results in 'a' being 'unknown'
const b = 0 || 'unknown' // results in 'b' being 'unknown'
const c = false || 'unknown' // results in 'c' being 'unknown'
const d = NaN || 'unknown' // results in 'd' being 'unknown'
```
This is because `''`, `0`, `false` and `NaN` are falsey values. The nullish coalescing operator however will only return the right-hand side operand if the left-hand side operand is either `null` or `undefined`.
## Practical example
Let's say you want to display `unknown` if somebody's amount of points is not known (arbitrary React.js sample code):
```jsx
const person = {
    name: 'Roberta Woods',
    email: 'roberta.woods@example.com',
    location: 'California',
    points: 0
};

return <dl>
    <dt>Name</dt>
    <dd>{person.name || 'unknown'}</dd>

    <dt>Email</dt>
    <dd>{person.email || 'unknown'}</dd>

    <dt>Location</dt>
    <dd>{person.location || 'unknown'}</dd>

    <dt>Points</dt>
    <dd>{person.points || 'unknown'}</dd>
</dl>;
```
In this case, `points` will display `unknown` which isn't right. We know the number of points, it's `0`. Previously, a fix would have been to also check if `points` is `0` or a number in general and then display that. By using the nullish coalescing operator we can keep our code clean and simple:
```jsx
const person = {
    name: 'Roberta Woods',
    email: 'roberta.woods@example.com',
    location: 'California',
    points: 0
};

return return <dl>
    <dt>Name</dt>
    <dd>{person.name || 'unknown'}</dd>

    <dt>Email</dt>
    <dd>{person.email || 'unknown'}</dd>

    <dt>Location</dt>
    <dd>{person.location || 'unknown'}</dd>

    <dt>Points</dt>
    <dd>{person.points ?? 'unknown'}</dd>
</dl>;
```
That's it! I won't bother you with a bunch of other examples. This feature is one of those additions that you will probably find helpful in a variety of code you write and need to check for values. As always, thanks for reading!

Nullish Coalescing (try to say it quickly ten times)

## A real-world use case
For one of the [Next.js]([https://nextjs.org/](https://nextjs.org/)) projects I was working on, I needed to add an asset prefix to the configuration if the [Node]([https://nodejs.org/en/](https://nodejs.org/en/)) environment wasn't "development". The first thing you might think of is to create a configuration object, write an if-statement and add a key to the configuration object if the statement is true. This would look something like this:
```js
require('dotenv').config();

const nextConfig = {
    poweredByHeader: false
};

if (process.env.NODE_ENV !== 'development' {
    nextConfig.assetPrefix = 'example.website';
}

export default nextConfig
```
In essence, we want to export an object with a few keys, one of which is optional. This is quite a bit of lines to do just that. I needed to declare a variable called `nextConfig` to optionally add a key to and then export it. This got me wondering if there is an easier way.
## The solution
The solution is quite elegant I think. This is what I came up with:
```js
require('dotenv').config();

export default {
    poweredByHeader: false,
    ...(process.env.NODE_ENV !== 'development' && { assetPrefix: 'example.website' }),
};
```
It might be a small decrease in lines of code, but imagine having multiple of these optional keys. In the first example, this would turn into quite a big file.

So Dave, why does this work? There are two parts to the solution. Firstly, I use the [Spread syntax]([https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Spread_syntax](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Spread_syntax)) to take all the keys from an object and add to to the configuration object like so:

```js
require('dotenv').config();

export default {
    poweredByHeader: false,
    ...{ assetPrefix: 'example.website' }
};

// result: { poweredByHeader: false, assetPrefix: 'example.website' }
```
I then add the statement to check whether the Node environment is "development" using `&&`. If it is not, it will try the spread the value `false` in the object. That falsey value will be omitted from the configuration object. This way it will optionally add the key to the configuration object.

There we have it, We have a sweet one-liner to optionally add a key to an object. Thanks for reading!

Optionally add a key to an object in one line

Writing (React.js) components can be very difficult. There are a lot of things to consider whil writing components for a project of any size. In this talk we take a look at some best practices for you and your team to follow. We do all of this while looking at code examples of an NPM drinking game!

ES6+

Articles

IndexedDB > Web Storage

You should use the Intl browser API

Quick Bits

Avoid the Async Await try-catch hell

Why I use data attributes over classes for declarative element states

Nullish Coalescing (try to say it quickly ten times)

Optionally add a key to an object in one line

Talks

N ways to write better React components for your next Nerdy Party Machine