UPDATE: Title has been changed because of the misunderstanding around it and the content. I hope this change will cover the expectations and keep healthy conversations to make us keep growing as developers, coders, and professionals in the end.
In my experience, more than 15 years of coding, programming, architecting, I've found people implementing code with no rules or standards and, sometimes, people believing they follow some rules but, in reality, they are not applying by themselves. I was in that situation a lot of times and keep being sometimes as well. I've written this article to show what I think is a good practice and makes us being good professionals.
The issue
The following React code renders a list of items in case of the array passed has them.
function List(props) {
const items = props.items;
return <ul>
{items && items.map(i => <li key={i}>{i}</li>)}
</ul>;
}
function App() {
const collection = [1, 2, 3, 4, 5, 6, 7, 8, 9, 0];
return <List items={collection} />;
}
ReactDOM.render(App(), document.getElementById('app'));
Do you think there is something wrong with the code? This code works perfectly fine, it creates a li
node with the number as the value.
What happens if the array has no items in it? The ul
node will be rendered as well but there won't be any li
node and no items in there. This is not a big issue but something not completely well.
We can modify the code in this way:
function List(props) {
const items = props.items;
// *** notice how the code is modified and complex increases
return items &&
<ul>
{ items.map(i => <li key={i}>{i}</li>) }
</ul>;
}
function App() {
const collection = [1, 2, 3, 4, 5, 6, 7, 8, 9, 0];
return <List items={collection} />;
}
ReactDOM.render(App(), document.getElementById('app'));
Like the previous case, this code is working fine, it does what it is supposed to do.
So, let me ask the same question as before, what happens if the array has no items in it?
In this case, a false
value and no HTML node are returned. The issue here is that we return different kinds of items depending on the items
property passed.
Why is this an issue? In fact, this is just a concept issue more than a coding issue. Returning the same kind of item in this function will make it easier for testing, make it easier to maintain, make it easier to read because it will be consistent and other methods calling this one will receive exactly what they expect and won't have the necessity to check if retrieves a boolean or a component.
The next code modification must have in mind this premise so it'll be like the following:
function List(props) {
const items = props.items;
// check if items is not empty
const isEmpty = !(items && items.length > 0);
// always return a React component
return isEmpty
? <React.Fragment />
: (<ul>
{ items.map(i => <li key={i}>{i}</li>) }
</ul>);
}
function App() {
const collection = [1, 2, 3, 4, 5, 6, 7, 8, 9, 0];
return <List items={collection} />;
}
ReactDOM.render(App(), document.getElementById('app'));
It seems that our code makes more sense now, doesn't it? We always return a component for any of the options or paths our code takes. If items is not empty, a ul
node is returned plus a li
node per item inside, a React component in fact.
If there is no items, a React.Fragment
component is returned. Both of them are React components, no need for callers to check it.
As a professional programmers that we are, we must give meaning and name our objects.
You may noticed we have few objects here in our code, an empty list component, an item list component, an item component and a manager component. Each of them with a unique responsibility (following the Single Responsibility principle that I'll talk in a future article) and a simple code easy to understand, to maintain and to test.
function ListItem(props) {
return <li>{props.value}</li>;
}
function ItemList(props) {
const items = props.items;
return <ul>
{ items.map(i => <ListItem key={i} value={i} />) }
</ul>;
}
// As suggested by Andy Nicholson
const EmptyList = () => <React.Fragment />;
function ListManager(props) {
const items = props.items;
const isEmpty = items && items.length <= 0;
return isEmpty
? <EmptyList />
: <ItemList items={items} />;
}
function App() {
const collection = [1, 2, 3, 4, 5, 6, 7, 8, 9, 0]
return <ListManager items={collection} />;
}
ReactDOM.render(App(), document.getElementById('app'));
I think this code looks like much better than the first one. Reviewers'll see the person behind the code thought how to face the problem, the constraints, and the paths to take around of it.
Wrapping up
As I wrote some paragraphs above, professional programmers must give meaning to the code. Coding can be done by anyone, programming in a good way with sense are just a subject of professionals.
What do you think about the exercise made in this article?
Hope this can be useful to you or just have fun reading it.
Top comments (26)
The title is clickbait and React.Fragment (
<></>
) has very valid use cases. The example you showed technically doesn't need to return a fragmentāyou can just doreturn !isEmpty && (...)
. Or just return null if it's empty, depending on what your preference is.There are other valid use cases for fragments, too, like if you need to return two siblings but without introducing unnecessary wrapper parents.
My apologies if you feel the title is clickbait. That wasn't my intention at all.
My article wants to point out that we should use the tools provided by frameworks and libraries in the scope of our domain solution. Returning
React.Fragment
as an empty list is something valid and will work but it doesn't mean nothing if we, as programmers, don't give it a meaning. Everything depends on requirements and, maybe, we don't need to deep dive in a big file structure with lots of components because it increase complexity.Your final code reads very well, thatās to be commended.
I donāt agree that the solution to every React problem is more components, though. Writing a ListManager component seems like massive overkill and poor naming: what you're describing is more like a ListItemManager - and what is a ListItemManager if not, well, a List?
Naming an empty list is great, but you donāt need a component for that either! You could do just as well by interpolating a const instead, saving yourself a function call:
React is awesome, but moar components arenāt always the answer.
Very good points.
IMO:
If your gonna return an empty fragment you might as well return
null
.And if your goal is readability, you can always use multiple
return
like thisThis is exactly what I came to the comments hoping to post, thanks for getting there first!
I'm baffled to see such a simple example turned into such a lot of code and it exemplifies the over-engineering the React community appears to deal with. Replace
null
with<React.Fragment />
if you must but creating a list shouldn't take 10s of lines of code.Multiple returns which both return the same type (html) is the most readable and reusable
I prefer this over returning fragment.
But in the most situations we need an empty state component instead of null
Null is more performant too. With null there's nothing for react to process, it just moves on. A fragment has to be processed, even if it isn't very much.
I much prefer this version:
The logic is inlined so you don't have to jump between component when reading the code.
You won't have ever new dev on your team asking you why you are returning and empty fragment.
The final example looks more "clean", but what actually happens is that you have to jump around in the file a lot to understand what is happening.
Agree with your last point. We should balance the benefits of creating components and get a complex solution from a file structure point of view.
And we need to stop promoting ideas like DRY and Single Responsibility Principle, which are at best pointless and at worst harmful.
I wish one day I'd be able to express my, points in such linear and organized way!
Good read! definitely scales me up!
Thanks @pracoon
Fragments were introduced as a workaround to JSX not supporting unwrapped sibling elements.
Professional programmers should probably start by reading the docs.
Full agree. Also, if the docs were read in this case, one would also see that the official advice is to return false ala
return condition && <li>item</li>
so that React immediately knows to not render anything. Not sure where the expectation that the component must return something renderable came from in this article.I'm confused š¤
Do you mean to say you're a coder but don't do it professionally?
Like, if you were a professional programmer you wouldn't have returned a fragment (since they don't do that sort of thing) but would have just returned
null
and been done with it?Anyway, I totally agree that to be effective as a coder you need to have all the tools in the toolkit at your disposal, and know the situations where they're useful.
To me, fragments' domain of usefulness is for encapsulating multiple elements, or unknown numbers of elements, as one return value. If I'm returning zero elements, I'll return
null
since that is whatnull
means: a typed but missing value.Fragment is very useful when you need to have a parent that does not need to render a container but need to render its children.
From example the Context Provider, component from react-router-dom.
I'm so confused. The title of the article says professionals never return "React.Fragment" but by the end of the article you have a component (EmptyList) that returns "React.Fragment" Does this mean you're not a professional?
In your first edit you have a comment to point out the added complexity, but the final edit has a ton of needless complexity, but is missing that comment.
Returning null for an empty list is perfectly testable, acceptable, and most importantly, simple. All things being equal, I'll opt for simplicity.
Totally agree with the requirements specification and domain model comments. My intention in this article is to show an example of how framework features have to be adapted to our solution and not in the opposite.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.