Ron Buckton
2014-10-16 18:11:24 UTC
??I recall from earlier discussions on this list that the reason `Set.prototype.add` returns `this` is to support chained calls to the set, to add multiple items, similar to how some libraries like jQuery operate, for example:
```
var s = new Set();
s.add(1).add(2).add(3);
```
I have found myself using Set more and more and have found that there are more instances where I would rather it return a boolean value, where `true` means the value was added to the set, and `false` when the value was not added as it already exists. Without this, I only have two options to detect whether an item was added to the set: (a) test using `Set.prototype.has` before adding the value, or (b) test using `Set.prototype.size` after adding the value.
# (a) Test using `Set.prototype.has`
```
var s = new Set();
...
if (!s.has(value)) {
s.add(value);
// code that executes only for unique values...
}
```
The problem with this solution is that we have to look up the value in the Set instance twice. As the Set grows, the cost of this algorithm will always be double the cost of the lookup. I imagine implementations can (and likely will) attempt to cache recent lookups for performance to mitigate this cost, however to a consumer of Set it could appear as if I'm still performing the same lookup twice.
# (b) Test using `Set.prototype.size`
```
var s = new Set();
...
var size = s.size;
s.add(value);
if (size < s.size) {
// code that executes only for unique values...
}
```
This solution removes the double lookup, but feels unnecessarily wordy and unintuitive.
# Proposal: `Set.prototype.add` returns Boolean
```
var s = new Set();
...
if (s.add(value)) {
// code that executes only for unique values
}
```
This seems the most concise and intuitive, and mirrors the `Set.prototype.delete` method's return behavior. From the consumer's perspective it has the appearance that the lookup only needs to be performed once.
Best regards,
Ron
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/es-discuss/attachments/20141016/36061f18/attachment-0001.html>
```
var s = new Set();
s.add(1).add(2).add(3);
```
I have found myself using Set more and more and have found that there are more instances where I would rather it return a boolean value, where `true` means the value was added to the set, and `false` when the value was not added as it already exists. Without this, I only have two options to detect whether an item was added to the set: (a) test using `Set.prototype.has` before adding the value, or (b) test using `Set.prototype.size` after adding the value.
# (a) Test using `Set.prototype.has`
```
var s = new Set();
...
if (!s.has(value)) {
s.add(value);
// code that executes only for unique values...
}
```
The problem with this solution is that we have to look up the value in the Set instance twice. As the Set grows, the cost of this algorithm will always be double the cost of the lookup. I imagine implementations can (and likely will) attempt to cache recent lookups for performance to mitigate this cost, however to a consumer of Set it could appear as if I'm still performing the same lookup twice.
# (b) Test using `Set.prototype.size`
```
var s = new Set();
...
var size = s.size;
s.add(value);
if (size < s.size) {
// code that executes only for unique values...
}
```
This solution removes the double lookup, but feels unnecessarily wordy and unintuitive.
# Proposal: `Set.prototype.add` returns Boolean
```
var s = new Set();
...
if (s.add(value)) {
// code that executes only for unique values
}
```
This seems the most concise and intuitive, and mirrors the `Set.prototype.delete` method's return behavior. From the consumer's perspective it has the appearance that the lookup only needs to be performed once.
Best regards,
Ron
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/es-discuss/attachments/20141016/36061f18/attachment-0001.html>