Mark Hammond
Mark Hammond
Sorry for no response, I just got back from a break! I also misunderstood the problem when I said "not particularly difficult" :) This will require some fiddling in that...
I see - although I guess I'm a little surprised you can't generate the bindings from the foreign target - maybe that's a better thing to lean in to? The...
> Are proc macros better suited for multi-crate interoperability? It's more like "multi-crate interoperability typically requires generating the bindings from a built library instead of from a single UDL". It...
Back to the issue :) This is quite difficult due to only the .rs generation needing to know the runtime. It might be easier to have an option to allow...
It looks like #1352 tried to do exactly this. I'm not sure if that regressed or was never implemented.
> Feel free to close this if it's not feasible. I had a bit of a look and it is feasible. It's a bit tricky because we can only do...
> only to discover it wouldn't be possible to use them at all in the end because for a generated interface FooInterface another BarInterface would have a method like ......
You can imagine metadata holding some named attributes for items - docstrings could also be carried here for example.
Hi, thanks for checking out UniFFI! UniFFI imposes many restrictions on your interface which means many Rust patterns do not really work. In particular, returning `&mut Self` isn't supported -...
> so maybe it makes sense to re-open this issue? Sure, let's do that!