Re: [RFC] Partial Function Application for instance of non-static methods ("$this")

On Thu, Jan 22, 2026, at 10:26 AM, Rowan Tommins [IMSoP] wrote: > On 22/01/2026 16:08, Tim Düsterhus wrote: >> Yes, it certainly requires some getting used to it. It's the best we >> could come up with, but if there are any better suggestions, we're >> open to hearing those. I'd like to note the “Rejected Features” >> section, since it lists some non-workable alternatives we considered. > > > Looking at this: > > > PHP itself already has "ALGOL-style" declarations, for "const", > "global", and "static". > > I'm reminded of placeholders in PostgreSQL; the type is needed to choose > overloaded functions and operators, so this doesn't compile (without > extra metadata): > > Select do_something(?) as result > > Instead, you can insert a dummy cast to indicate the type: > > Select do_something(?::int) as result > > > The PHP equivalent would be something like this: > > $f = (DateTimeImmutable)?->format('c'); > > Or, slightly more readable but less consistent with casts: > > $f = (? as DateTimeImmutable)->format('c'); > > > That doesn't solve the second part of the problem though: > > > It would also not allow to reorder the parameters in the resulting > Closure, which is an explicit feature of Partial Function Application. > > > I thought I'd throw it out there anyway in case it stimulates any other > ideas. More spitballing on my previous reply: class Test { public function stuff(int $a) {} } (Test)?->stuff(?); Translates to: fn(Test $__this, int $a) => $__this->stuff($a); But if there's no partialing on the method side, it could get abbreviated to: (?)->stuff(4); Translates to: fn(object $__this) => $__this->stuff(4); Because we don't need to know the type at compile time, and a method-not-found error would still happen at runtime anyway if necessary. That would then be a lot easier to write in cases where you're just dropping a $this->stuff() call into a pipe chain but want to receive $this. I'm not sure if that would also allow for reading properties with the same syntax? --Larry Garfield