auto_impl for trait objects? about auto_impl HOT 3 CLOSED

It would be absolutely possible to extend auto_impl for this use case! But at the same time, we should also support &dyn Trait, &mut dyn Trait, Rc<dyn Trait> and Arc<dyn Trait>. So basically every "pointer type" we currently support, only with a trait object.

I'm not quite sure how much work that would be. Hopefully we don't have to copy all the code for the other types, but can instead abstract somehow. Also the Send, Sync and lifetime bounds could be complicated. Mhhh...

Maybe it could be worth it to rewrite much of what we have: instead of hand coding each type we support, we define a table where we list each type and what restrictions it has. Then our system would be super generic and adding types would be easy.

I won't have time in the next months to implement this, but I could help you out if you want to try it!

from auto_impl.

I had a poke around the source and came to similar conclusions. I was thinking of adding it as something like #[auto_traitobj(&, Rc, Arc + Sync + 'static)] - ie, basically the same syntax but with the ability to append marker traits.

As far as codegen goes, it would be nice to avoid duplicating all the existing code, but I haven't looked at it closely enough yet to work out how to do that.

from auto_impl.

#52 addresses this more precisely.

from auto_impl.