RInno makes it easy to install local shiny apps by providing an interface between R, Inno Setup, an installer for Windows programs (sorry Mac and Linux users), and Electron, a modern desktop framework used by companies like Github, Slack, Microsoft, Facebook and Docker. RInno is designed to be simple to use (two lines of code at a minimum), yet comprehensive.
If a user does not have R installed, the RInno installer can be
configured to ask them to install R along with a shiny app,
include_R = TRUE
. And similar to Dr. Lee Pang’s
DesktopDeployR project,
RInno provides a framework for managing software dependencies and error
logging features. However, RInno also supports GitHub package
dependencies, continuous installation (auto-update on start up), and it
is easier to manage with create_app
, the main RInno function.
DesktopDeployR requires many manual adjustments and a deep understanding
of the entire framework to use, but RInno can be learned incrementally
and changes automatically flow down stream. You don’t need to remember
the 100+ places impacted by changing app_dir
. RInno only requires a
high-level understanding of what you’d like to accomplish.
# Get remotes package
install.packages("remotes"); require(remotes)
# Use install_github to get RInno
install_github("ficonsulting/RInno")
# Require Package
require(RInno)
# Use RInno to get Inno Setup
install_inno()
Once you have developed a shiny app, you can build an installer with
create_app
followed by compile_iss
.
# Example app included with RInno package
example_app(app_dir = "app")
# Build an installer
create_app(app_name = "Your appname", app_dir = "app")
compile_iss()
create_app
creates an installation framework in your app’s directory,
app_dir
. The main components are a file called “app_name.iss” and the
“nativefier-app” directory. You can perform minor customizations before
you call compile_iss
. For example, you can replace the default/setup
icon at Flaticon.com, or you can customize
the pre-/post- install messages, infobefore.txt and infoafter.txt.
Just remember, the default values
(i.e. create_app(info_after = "infobefore.txt")
) for those files have
not changed. The Inno Setup Script (ISS), app_name.iss, will look for
default.ico and try to use it until you update the script or call
create_app
with the new icon’s file name
(i.e. create_app(app_icon = "new.ico")
). Likewise, the Electron app
will need to be recompiled to capture any manual changes to files in
app_dir
.
Electron is now used to render the shiny app’s UI. All other
user_browser
options will be deprecated in future releases.
In order to replace Electron’s logo with your app’s icon, add something like this to your ui.R file:
fluidPage(
tags$head(
tags$link(
rel = "icon",
type = "image/x-icon",
href = "http://localhost:1984/default.ico")
)
)
In order to close the app when your user’s session completes:
- Add
session
to yourserver
function - Call
stopApp()
when the session ends
function(input, output, session) {
if (!interactive()) {
session$onSessionEnded(function() {
stopApp()
q("no")
})
}
}
If you forget to do this, users will complain that their icons are broken and rightly blame you for it (an R session will be running in the background hosting the app, but they will need to press ctrl + alt + delete and use their task manager to close it). Not cool.
Provide a named character vector of packages to create_app
, and RInno
will download them and install them with your shiny app. RInno downloads
windows binaries from CRAN for the listed packages and their
dependencies with
tools::package_dependencies(packages = pkgs, recursive = TRUE)
.
create_app(
app_name = "myapp",
app_dir = "app",
pkgs = c("shiny", "jsonlite", "httr")
)
For remotes
, Github source files are compiled into windows binaries.
Bitbucket will be supported in a future release.
If you would like to create a custom installer from within R, you can
slowly build up to it with create_app
, like this:
create_app(
app_name = "My AppName",
app_dir = "My/app/path",
dir_out = "wizard",
pkgs = c("jsonlite", "shiny", "magrittr", "xkcd"), # CRAN-like repo packages
remotes = c("talgalili/installr", "daattali/shinyjs"), # GitHub packages
include_R = TRUE, # Download R and install it with your app, if necessary
R_version = "2.2.1", # Old versions of R
privilege = "high", # Admin only installation
default_dir = "pf") # Install app in to Program Files
create_app
passes its arguments to most of the other support functions
in RInno. You can (and probably should) specify most things there and
they will get passed on. Alternatively, you can provide instructions
directly to those support functions like this:
# Copy installation scripts (JavaScript, icons, infobefore.txt, package_manager.R, launch_app.R)
copy_installation(app_dir = "my/app/path")
# If your users need R installed:
get_R(app_dir = "my/app/path", R_version = "2.2.1")
# Create batch file
create_bat(app_name = "My AppName", app_dir = "my/app/path")
# Create app config file
create_config(app_name = "My AppName", R_version = "2.2.1", app_dir = "my/app/path",
pkgs = c("jsonlite", "shiny", "magrittr", "dplyr", "caret", "xkcd"))
# Build the iss script
start_iss(app_name = "My AppName") %>%
# C-like directives
directives_section(R_version = "2.2.1",
include_R = TRUE,
app_version = "0.1.2",
publisher = "Your Company",
main_url = "yourcompany.com") %>%
# Setup Section
setup_section(output_dir = "wizard",
app_version = "0.1.2",
default_dir = "pf",
privilege = "high",
inst_readme = "pre-install instructions.txt",
setup_icon = "myicon.ico",
pub_url = "mycompany.com",
sup_url = "mycompany.github.com/issues",
upd_url = "mycompany.github.com") %>%
# Languages Section
languages_section() %>%
# Tasks Section
tasks_section(desktop_icon = FALSE) %>%
# Files Section
files_section(app_dir = "my/app/path", file_list = "path/to/extra/files") %>%
# Icons Section
icons_section(app_desc = "This is my local shiny app",
app_icon = "notdefault.ico",
prog_menu_icon = FALSE,
desktop_icon = FALSE) %>%
# Execution & Pascal code to check registry during installation
# If the user has R, don't give them an extra copy
# If the user needs R, give it to them
run_section() %>%
code_section() %>%
# Write the Inno Setup script
writeLines(file.path("my/app/path", "My AppName.iss"))
# Check your files, then
compile_iss()
Feel free to read the Inno Setup documentation and RInno’s documentation to get a sense for what is possible. Also, please suggest useful features or build them yourself! We have a very positive culture at FI Consulting, and we would love to get your feedback.
Please note that this project has a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.
The RInno package is licensed under the GPLv3. See LICENSE for details.