Re: [RFC] Readonly property hooks

On Tue, Jul 22, 2025, at 11:18, Claude Pache wrote: > > >> Le 22 juil. 2025 à 09:43, Rob Landers <rob@bottled.codes> a écrit : >> >> You can also implement this today, without using readonly: >> >> class Foo { >> public int $bar { >> get => $this->bar; >> set => empty($this->bar) ? $this->bar = $value : throw new >> LogicException("nope"); >> } >> } >> >> https://3v4l.org/2JagR#v8.4.10 >> > > Your code is buggy, because 0 is “empty”: > > https://3v4l.org/iUPvW#v8.4.10 I wrote it over 30s while getting ready to walk out the door for a week long vacation, so I won’t be entirely surprised if it is wrong. :) > >> Should I be able to mark this class as readonly? I would think so. > > I don’t think so. > > If you want to *document* the intended invariant, you can put a @readonly tag in a phpdoc > comment. > > Adding a readonly keyword should *enforce* the invariant; the > added value is that it would choke on bugs like the one you wrote just above, making it debugging > much easier. I’m not sure if you meant to, but I feel like you just argued for allowing readonly on hooks so that these kinds of bugs aren’t accidentally written… > > >> The readonly keyword simplified that greatly, however, readonly has been neutered compared >> to regular classes in the last couple of versions. There are so many edge cases and non-implemented >> features with them -- mostly due to this exact argument you are making -- that they're nearly >> worthless to actually define immutable value objects in today's PHP. >> > > There are several issues with the readonly feature, mostly because of the “worse-is-better” > philosophy. But we can slowly correct the main issues instead of making them worse. > > —Claude I think this is the main crux of the issue, right? There is one camp that says readonly means immutable and another that says readonly is read-only. These two viewpoints are not compatible despite having a lot of overlap. — Rob