When in Doubt, Upgrade | /en/2017/05/when-in-doubt-upgrade/

Is installr no longer Windows only?

Upgrading R only needs to be done once in several months, so it is not too bad to download the installer from CRAN and install it manually on Windows. I think R_LIBS_USER=~/R plus update.packages(checkBuilt = TRUE) is a more fundamental and portable (in terms of operating systems) solution to avoid headache in updating R packages after upgrading R.

If we only talk about Windows, I feel Scoop ( http://scoop.sh ) is a more general solution to installing software on Windows. The idea is similar to Homebrew on macOS. In theory, all you need to do will be scoop update r. Just like I believe Homebrew is the future for macOS R, I believe scoop could be the future of both R and Rtools on Windows, but there is definitely still a long way to go.

Originally posted on 2017-05-19 18:41:37

Guest *Tal Galili* @ 2017-05-19 22:18:33 originally posted:

I hope for the day that installr would no longer be relevant. But I suspect that Windows users are often of the demographic of less tech savvy people than linux people (just a hunch, obviously).
Hence, offering people to use from their console simply:

install.packages("installr")
installr::updateR()

Appears to be of use to people (hence, I suspect, the 386K downloads thus far of the package: https://github.com/talgalili/installr/ )

As for non Windows solutions, there is the following first-stab solution for mac OS:
https://github.com/AndreaCirilloAC/updateR
I offered the people their to merge our projects into installr, and they seem to be positive about the notion (AndreaCirilloAC/updateR#10).
If you have suggestions, I'm sure your input would be super helpful.

(p.s.: installr was my first R package, so their is code there that I would probably not approve of these days. So if you get to have a look, please be your usual kind self :) )

Cheers,
Tal

Guest *Aaron Simumba* @ 2020-04-25 07:00:35 originally posted:

Every time I want to update my R installation. The suggestion by Yihui always come to mind. I have a directory set for my packages in the .Renviron.

I remembered this post while going through upgrading to R 4.0.0.

Actually I also pointed this post to a few other people yesterday because they had exactly the same issue. With R_LIBS_USER in .Renviron, and update.packages(ask = FALSE, checkBuilt = TRUE), there is nothing to worry about packages installed in the previous version of R.

Originally posted on 2020-04-25 15:10:23

Guest *Aaron Simumba* @ 2020-04-26 13:24:50 originally posted:

Nothing more nor less to complicate life. I choose to watch R do what it does best, that is update and upgrade my packages while I catch a breather...

Search for "remote" on this page: https://yihui.org/tinytex/

Originally posted on 2020-09-06 01:52:47