Davis Vaughan

I haven't opened any of these yet, if you want to do one for dbplyr that'd be great, and then I can take a look if you want

For what it's worth, I rewrote most of `tibbletime` to use `sloop::reconstruct()`. It seems to work for most cases, but I also needed a `maybe_reconstruct()` function that just returns a...

I think you're right that I could put the checks inside `reconstruct()` rather than include a separate function. They are cheap and quick checks so it wouldn't be a problem....

I think your suggestion removes a lot of unnecessary work on my end. I really appreciate you taking a look at it. I saw your latest PR, removing `...` in...

I'm actually surprised he doesn't do it there. In theory `new_Date()` should (I think?) be extensible so people could build on top of it. `new_Date()` and `new_s3_dbl()` are both in...

If we are being _really_ picky, the tibble specific attributes go before the `...` (see `new_my_class()` above). Just thought I'd point it out! ```r function (x, ..., nrow = NULL,...

https://github.com/tidyverse/tidyr/blob/944f5ad2ce3e1ab8148becb0e143a053b5a7aa14/R/complete.R#L83

I _think_ this is probably correct reasoning, as it gives S3 classes built on top of types that would otherwise be unsupported a chance to be converted to either: -...

As mentioned in r-lib/vroom#391, `check_column_types()` is currently a little broken because it doesn't correctly check for list columns, so we should fix that too.

Looking at how `dplyr_reconstruct()` is used in dplyr, it seems like it would be possible to have a single generic, `tibble_reconstruct()`, that dplyr uses instead of `dplyr_reconstruct()`. I'm not sure...