Re: [php-src] PR #20903: GH-10497 (const obj->prop write) - language change notes

> Hi, > The RFC talks only about global constants. What about class constants? It is somewhat > cumbersome to get a mutable object in this position, but it is nevertheless possible, and, > at least for consistency, it should be ensured that they are treated the same way as global constants: > https://3v4l.org/QhM1L Hi Claude, Good point, thanks. I've extended the implementation so they're handled the same way; both forms from your example now work: c::o->b = 42; $c::o->b = 42; One consequence worth flagging: enum cases are class constants, so unset(Enum::Case->value) (and writes to readonly case properties) now raise the regular runtime readonly-property Error instead of the old compile-time "temporary expression" error, i.e. identical to $x = Enum::Case; unset($x->value);. Enum case properties remain immutable. I've updated the PR (#20903) and bumped the RFC <https://wiki.php.net/rfc/const_object_property_write> to v0.3 to cover this. Since it widens the scope, I'll restart the discussion cooldown and push the vote date back accordingly. Best regards, __ *Khaled Alam* On Mon, Jun 22, 2026 at 12:01 AM Claude Pache <claude.pache@gmail.com> wrote: > > > Le 21 juin 2026 à 20:33, Khaled Alam <khaledalam.net@gmail.com> a écrit : > >  > Hi all, > > Reviving this thread. The v0.2 RFC > <https://wiki.php.net/rfc/const_object_property_write> > (object property > writes on constants only; > array/dim writes removed) addressed Tim's concerns, and the 14-day > cooldown has long passed. > > Unless there are further objections, I intend to open voting on Monday, 29 > June 2026. 2/3 majority, > targets PHP 8.6, two-week vote. > > RFC: https://wiki.php.net/rfc/const_object_property_write > PR: https://github.com/php/php-src/pull/20903 > > Best regards, > __ > *Khaled Alam* > > > Hi, > > The RFC talks only about global constants. What about class constants? It > is somewhat cumbersome to get a mutable object in this position, but it is > nevertheless possible, and, at least for consistency, it should be ensured > that they are treated the same way as global constants: > > https://3v4l.org/QhM1L > > > > —Claude > >