Hi all,
> When I was working on the readonly class RFC, I realized that the readonly > keyword very naturally fits services besides value objects. So my > expectation has been that until we can fix the issue with cloning, people > would mainly apply readonly to services. Not that it is very useful, but I > would also feel some kind of weird fulfillment by doing so. > > Regarding cloning: I created a WIP PR not long ago to fix the > aforementioned cloning issue, and I'll pursue a readonly amendment RFC in > the coming weeks (or month) containing the long awaited improvements for > cloning (hopefully together with the "clone with" construct) and possibly > with this inheritance-related change Nicolas proposed, unless someone can > come up with an ultimate counter-argument. > What's your take about 8.2? As I demonstrated, readonly classes are broken because of this propagation to child classes. Does that mean we should remove this constraint from 8.2? What about reverting the feature and considering it again for 8.3, after fixing it? I'm looking forward to your work on cloning! Note that "clone-with" or similar is a separate issue from the current inability to clone readonly properties within the __clone method. It might be easier to tackle them independently, from a discussion pov. Cheers, Nicolas
