Dealing with common, frequently recurring problems on update

[andrewH]

I have been using R for over a decade, and every time I update my packages -- every time without exception -- I get at least four fundamentally different warnings, error messages, or other statements that some package could not be updated. It usually takes me a good half-day to find all these problems, track down candidate solutions, test them, and implement the ones that work. I always change my code to eliminate known problems if I can.

I have also observed that more specific versions of this question are very common on stackoverflow.

My question is this: Has anyone put together a troubleshooting guide for problems of this sort, listing the more common problems, how to recognize each, and what to do about them? I am especially interested in learning of common configuration problems that can lead to multiple error types. I don't see how I can have this many different errors after running almost the same code dozens if not hundreds of times, but I do.

I always run everything as administrator. Pretty much all my packages have now been installed in two different places, C:\Program Files\R\R-4.1.2 and D:/Documents/R/win-library/4.0, and I don't know why. The first line of every program sets .libPaths to the two paths listed. Note that my documents directory has been remapped to the latter path for every purpose.

This is a Windows 10 box, and I've been using windows machines for over 40 years, but I have no formal training in computer science or computer administration (my background is in law, economics, and public policy), and no IT support.

It would be an amazing joy to be able to run update.packages(), go out for a cup of coffee, come back, and find my packages updated. I have literally (and I mean literally literally) never once had that happen.

[technocrat]

Yeah, I'm a lawyer, too, and this kind of stuff is not how I want to be spending my time. Unfortunately, I stopped using Windows with XP's endless parade of patches and went back to *nix (macOS for a long time and now Ubuntu). If I were in your position, I'd consider running a virtual Ubuntu for R work. The range of potential issues is much narrower, usually limited to installing some library dependency, which is easy to track down from the CRAN description page. It's rare for an update to have a missing dependency unless it is a full point upgrade.

If that's not in the cards, there are lots of Winfolk here, but you probably will get better help by posting your four types of common error and asking the question about cheats.

[andrewH]

< sigh > Yes, I periodically tell myself I should switch over to some flavor of -nix. It is a lot more compatible with my ethical beliefs about property that is expensive to produce but nearly free to reproduce. Bht when I think of having to go through all the hours I have spent figuring out how Windows works again, I admit that I quail. Plus I have enough trouble remembering how to do things with all the cues provided by a GUI. Even a greatly simlified text-based interface still requires that I actually know the commands, and don't just stare at the screen until I find something that might be relevant.