Jim Hester

One alternative would be to call `format(trim = TRUE, justify = "none")` which should turn off padding in all cases. ``` r format(c(1, 10, 100, 1000), trim = TRUE, justify...

Those are all characters, how is this not type stable?

Ah I see, I think 0 length inputs generating 0 length outputs is the correct behavior, but can see how it makes this particular idiom more painful.

Yeah this is similar to https://github.com/tidyverse/glue/issues/104

I agree this is a useful thing to check, however I am not sure devtools is the right place for it. It seems more appropriate as a [lintr](https://github.com/jimhester/lintr) linter or...

hours greater than 24 hours don't seem like valid datetimes to me, I don't think this will change.

Can you link to documentation that describes this, I don't understand it.

Sure, we can move to using the v4 endpoint. I am not sure we need to worry about adding more CI services, what we have seem sufficient for the vast...

I am pretty sure this is due to processx using SIGKILL in a number of places and the finalizer that is writing the coverage is interrupted. @gaborcsardi would know better...

Sure, but if the process is writing the coverage out in an exit finalizer the process is already in the process of exiting, so there must be some case where...