Fix donttest (#7)

* replacing donttest with dontrun

* version number to dev (and some trailing spaces)

* getting ready for release

- update version to 0.5.1
- add new change to news
- update cran comments to reflect current update

* updating the docs for the changed example

* updating cran comments to reflect R versions tested

DESCRIPTION

@@ -1,14 +1,14 @@
 Package: glmmboot
 Type: Package
 Title: Bootstrap Resampling for Mixed Effects and Plain Models
-Version: 0.5.0
+Version: 0.5.1
 Authors@R: person("Colman", "Humphrey", email = "[email protected]",
   role = c("aut", "cre"))
 Description: Performs bootstrap resampling for most models that update() works for. There
     are two primary functions: bootstrap_model() performs block resampling if random effects
     are present, and case resampling if not; bootstrap_ci() converts output from
     bootstrap model runs into confidence intervals and p-values. By default,
-    bootstrap_model() calls bootstrap_ci(). 
+    bootstrap_model() calls bootstrap_ci().
     Package motivated by Humphrey and Swingley (2018) <arXiv:1805.08670>.
 License: AGPL-3 | file LICENSE
 URL: https://github.com/ColmanHumphrey/glmmboot
@@ -18,7 +18,7 @@ Depends:
   R (>= 3.1)
 Imports: methods,
     stats
-Suggests: 
+Suggests:
     glmmTMB (>= 0.2.1),
     testthat (>= 0.11.0),
     parallel (>= 3.0.0),

NEWS.md

@@ -1,3 +1,7 @@
+# glmmboot 0.5.1
+
+* Changing a `donttest` to a `dontrun` within an example.
+
 # glmmboot 0.5.0
 
 * if `parallelism = "parallel"` but `num_cores` is left as NULL, we default to using `parallel::detectCores() - 1L` cores. This was in fact the documented behaviour previously, just sadly not the actual behaviour
@@ -26,14 +30,11 @@ API changes:
 * Added a warning if data not supplied explicitly
 * Changed two-sided p-value calculation to more tightly respect the fact that the bootstrap t-values
   intentionally are not forced to be symmetric.
-  
+
 # Travis CI:
 [![Travis-CI Build Status](https://travis-ci.org/ColmanHumphrey/glmmboot.svg?branch=master)](https://travis-ci.org/ColmanHumphrey/glmmboot)
 
 # glmmboot 0.1.2
 
 * Added a `NEWS.md` file to track changes to the package.
 * Initial release of glmmboot
-
-
-

R/bootstrap_methods.R

@@ -85,7 +85,7 @@ gen_samp_lev <- function(levels,
 #' @return Returns a vector of indices from orig_vector that correspond to
 #'   values in new_vector.
 #' @examples
-#' \donttest{
+#' \dontrun{
 #'     orig_vector <- c(1, 1, 2, 3, 3, 3)
 #'     new_vector <- c(1, 2, 1)
 #'

cran-comments.md

@@ -1,18 +1,11 @@
 ## Release Summary
 
-As requested, fixes issues where `class()` was assumed to return a length-one result (mostly was used in testing). All should be done now.
-
-Apart from that, primary change is to add a better default for `parallelism = "parallel"`, and to provide a solve for issues with `parallelism = "future"`:
-
-* if `parallelism = "parallel"` but `num_cores` is left as NULL, we default to using `parallel::detectCores() - 1L` cores. This was in fact the documented behaviour previously, just sadly not the actual behaviour
-* `bootstrap_model` (and internal function `bootstrap_runner`) accepts an argument `future_packages` that will be passed along to `future.apply::future.lapply`; this is needed for futures that don't share memory, because the required global isn't visible (when using S3 generics). This is also passed along to README/vignettes/tests etc.
-
-Some very minor documentation and messaging edits rounds out the submission.
+Changing `donttest` to `dontrun` for an example.
 
 ## Test environments
 * local OS X install, R 3.6.2
-* Ubuntu 16.04.6 LTS (on travis-ci), R 3.6.1
-* win-builder (devel and release), R 3.6.1
+* Ubuntu 16.04.6 LTS (on travis-ci), R 3.6.2
+* win-builder, devel R 4.0.0 alpha, and release R 3.6.3
 
 ## R CMD check results