Le 10/07/2026 à 20:33, Holly Schilling a écrit :
.
As far as scalar methods goes, I hate my implementation. I’m not suggesting
someone else could do better, only that the compromises necessary to implement
it weren’t worth it. After having implemented it and seen what it demands, I
think the right choice is Autoboxing with extensions on the box rather than
ever allowing `$this` to not be an object.
This is a heated topic lately and I don’t pretend to be an expert.
Typing this email this morning gave me real hesitation. If I can’t support my
own implementation, no one else should either. I immediately set out to build a
better version that I could put my full weight behind.
Autoboxing turned out to be a non-starter as well. The code just didn’t fit the
model of the code PHP uses. That led me to a diffident path.
C# 14 recently added a new extensions syntax to allow for properties and more.
When I saw it, I knew it was the right solution.
```
extension string $str {
function length() {
return strlen($str);
}
}
var_dump(“test”->length()); // int(4)
```
By declaring the parameter inline with the extension declaration, we eliminate
the downcast of `$this`, the fake accessibility constraints, and all the sharp
edges the old way introduces.
I immediately rewrote all 3 RFCs to use this new syntax, as well as the code to
implement it.
Anyone who has been following these discussions, please skim the updated docs
for changes.
Thanks for these changes, they bring better stuff to the table indeed :)
Still my first thought stands: I think scalar extensions are
dispensible, and having scalars behave "like objects" is quite new to
PHP workarounds do exist for that: there are quite a few PHP packages to
manipulate strings, utf8 handling, graphemes or codepoints, etc., that
could be used "as string extensions" if such RFC came to become real in
the language. I think it needs more time, investigation on packages,
whether to decide on adding "standard extensions" to start uniformizing
string-related functions (something that's been asked for a long time).
There's also a few libraries that could benefit from it, like
nette/utils, with their String manipulation API (check it out:
https://doc.nette.org/en/utils/strings).
For short: great idea, but needs more benchmarks, packages and ecosystem
analyses, etc.
