Re: [RFC] Readonly property hooks

On Fri, Jul 18, 2025, at 9:50 AM, Marco Pivetta wrote: > On Fri, 18 Jul 2025 at 16:43, Tim Düsterhus <tim@bastelstu.be> wrote: >> On 7/18/25 16:25, Faizan Akram Dar wrote: >> > The problem with allowing only set hooks is that readonly class won't be >> > compatible with hooks, I think that is one of the main motivations behind >> > this RFC. >> >> Yes. The point is that the semantics users expect from >> readonly are >> fundamentally incompatible with a get hook that could return arbitrarily >> changing values whenever you read from a property. > > Just a heads up: I also plan to vote "no" on this RFC because the > expectation with readonly is that there is no kind of > interference or > lazy initialization anyway. > > Now that lazy proxies have landed into core, there is also no need for > __get hacks anymore. Even on set hooks, which do not violate any of the interpretations of what readonly "means" floating about? As the RFC notes, that's now a necessary validation step with the improved clone() on its way. --Larry Garfield