Jim Hester

partially named lists are esoteric enough in practice that I am not sure it is worth the added complexity. But I don't have a very strong opinion in either direction.

Yes, I think it is a good idea

I agree, having a more informative error during registration would help a lot.

It is not clear from the above what is 'inaccurate'. Can you provide a **minimal** [reproducible example](https://www.tidyverse.org/help/#reprex) that illustrates the issue?

Thank you for the reproducible example, however it is certainly not minimal. `bench::bench_memory()` is a very minimal wrapper around `utils::Rprofmem()` I doubt the issue lies there. I think likely lazy...

I think probably cli should set `comment = character()` when calling glue to work around the `#` issue. For the unpaired quote issues I guess we would need to add...

Thanks for working on this, but I am confident that the symbols will work regardless of the R version and we use symbols for packages for a reason. I suspect...

Sure we can expose the argument, though in general I don't think most packages would ever need to use it. Also FWIW if you use `usethis::use_cpp11()` to setup cpp11 for...

For package use the cpp11 namespace won't exist, so the namespace of the package using cpp11 is probably the best option as Lionel suggested. If you are using `cpp11::source_cpp()` and...

I think this makes sense as a way for packages to extend the conversions as needed. If you and @romainfrancois think using the conversion trait struct is easier to understand...