Davis Vaughan comments

Results 580 comments of


                                            Davis Vaughan

Update test for `problems()` `unwind::protect` error

See https://github.com/r-lib/cpp11/issues/274 for full details of the crash

PERFORMANCE: Search large `X`:s for globals is slow [SOLVED]

This is definitely helpful. I was just doing some testing iterating over `rsample` cross-validation objects (lists containing data frames and other structures deep inside). Before the fix, searching `X` was...

Overhead of future_lapply relative to parLapply/mclapply, etc.

@HenrikBengtsson FWIW you can inject `NULL` into an expression (or list) with `[` (not `[[`!) and `list(NULL)` ``` r x

Creating a progressor object outside `with_progress()`?

Upon more study, this works with `initiate = FALSE`. It requires setting `steps` (which is reasonable), but also requires setting `auto_finish`. I wonder if that could default to `TRUE` somewhere...

Creating a progressor object outside `with_progress()`?

I think you are right, I was just thinking about this example below from the perspective of a user. When writing `slow_fn()`, I'd be "using" `p()` before I ever create...

Creating a progressor object outside `with_progress()`?

Oh actually I don't think that creating `slow_fn()` ahead of time is a viable solution. The following example won't work at all (ignoring the fact that `with_progress()` currently returns `NULL`,...

Creating a progressor object outside `with_progress()`?

> I think the proper way to handle this by passing the progressor function via an argument, e.g. The more I think about all this, the more I think that...

Argument roll=FALSE is not working as expected in update

With the clock approach, you generally only update one component at a time, which forces you to resolve the _invalid date_ as soon as it arises. ``` r library(clock) library(magrittr)...

Add `as.POSIXct()` and `as.Date()` methods for Period objects that simply error

Hmm, possibly? Might be a hard thing to sell to R Core. In the meantime it probably wouldn't hurt to add our own methods (and we could probably give a...

Version mismatch between CRAN and the website and changelog

The "auto" development mode that is set [here](https://github.com/tidyverse/lubridate/blob/c2e4c1d994eaad7a921fc89d503d1bdc1d73ea05/_pkgdown.yml#L6) will push the docs to the normal release site when the package version number is a non-dev version (i.e. it doesn't contain...