pkgload
pkgload copied to clipboard
r-lib
Simulate installing and loading a package
So that rlang does not need to provide `env_unlock()` which is implemented with non-API calls. Probably need to remove `reset = FALSE` to achieve this?
Used in RStudio already: https://github.com/rstudio/rstudio/blob/4a20e65e129178ecadcffc53b7ad48b2c9781769/src/cpp/session/modules/SessionHelp.R#L781-L792
I see the following (truncated) output for `pkgload::load_all("Matrix")` for the package cloned from the `cran/Matrix` GitHub repo: ```r ─ DONE (Matrix) Error in readChar(con, 5L, useBytes = TRUE) : cannot...
Thanks for the package, it is very useful! As I was developing a package, I encountered some issues with `roxygen2` which I tracked down to `pkgload::unregister`. When unregistering S3 methods,...
I got into a situation that can be simplified like this : ``` # in development with devtools old_ns ℹ Loading mypkg new_ns new_ns #> identical(old_ns, new_ns) #> [1] FALSE...
Followup to #269. Adds cross-reference to `is_dev_package()` from `load_all()` documentation.
Probably because of the changes in 2d28e266bf111b3e49a5e1a473b053c7104d5459 where we no longer call `unloadNamespace()`. So now we should call `.onUnload()` if present, and similarly handle the `onUnload` hook.
To re produce, within any package project (named PACKAGE below) : ``` devtools::load_all(".") #> ℹ Loading PACKAGE search() #> [1] ".GlobalEnv" "devtools_shims" "package:PACKAGE" "package:testthat" "tools:rstudio" "package:stats" #> [7] "package:graphics" "package:grDevices"...
Hello, I want to report the following bug / weird behaviour: 1. I do `pkgload::load_all()` once. 2. Then I change something in any file. Just a small change. 3. Then...
I need a manual `unloadNamespace()`, before or after the second `load_all()`, to be actually able to use it. Should we unload all dependent namespaces when `load_all()` is reloading? ``` r...
