Makes sure save_image() can start Chrome, without asking: finds a
Chrome or Chromium (see save_image()) or else downloads Chrome Headless
Shell with chromote, and on Linux downloads the shared libraries it needs
that are missing. The libraries come from the system's own package
repositories (with apt-get download or dnf download, which need no
administrator rights) and are unpacked into plotly.core's user cache
(tools::R_user_dir("plotly.core", "cache")), where save_image() finds
them. Run it once on a new machine, e.g. before scripts that export images;
interactive sessions are offered the same downloads when needed.
Details
Downloaded libraries are only used for what the system lacks: libraries
installed on the system later (and its updates) take over. They are kept in
a folder for the OS release, and those for other releases (e.g. after an
upgrade) are removed when libraries are next downloaded. They don't get the
system's updates, so once they are more than
getOption("plotly.core.chrome_libs_max_age", 90) days old, or Chrome has
changed, install_chrome() downloads them again (interactive sessions are
asked; scripts get a message saying to run install_chrome()). remove_chrome() removes everything plotly.core has
downloaded and, with browser = TRUE, the Chrome Headless Shell builds
chromote downloaded.
The downloaded software is not part of plotly.core and keeps its own
licence: Chrome Headless Shell is built from Chromium by Google (see the
licence files in its download), and the libraries are your distribution's
packages (each package's copyright file is kept with it, under
usr/share/doc).