Jim Hester

The big benefit to using `foo

There is a bit of a chicken and egg problem here, standardising column types needs column names sorted out first, but if column names depend on skipped columns ☹️ ....

There isn't really enough detail here for me to be sure what you are talking about. But if it is `delayedAssign()` this is such a rarely needed technique I am...

Base R does not support writing to zip files from a connection. e.g. there is no write equivalent to `unz()`. However you can now use the [archive package](https://archive.r-lib.org/) to do...

yeah, I think that would make sense.

The following works for the test case, but generates a warning and incorrect results when doing arithmetic with Date objects. ``` r Ops.hms [1] "2016-05-30" ```

I agree, the current behavior is not ideal, I agree only showing this dialog once would be ideal. PRs welcome!

This is likely a network failure that never times out, probably from homebrew, I don't think there is anything we can do, just cancel these jobs if you see one...

Agreed that would be nice, looks good! The vroom benchmarks actually aren't using `scale_x_bench_time`, they are just using `scale_y_continuous(labels = function(x) format(bench::as_bench_time(x)))`, but nicer breaks would definitely be nice!

My main hesitation is that currently the read only vector classes never allocate on Rs heap, which seems a nice property when you are reasoning about them. Really we should...