R：在更新R或RStudio之后更新或安装软件包存在许多问题

Question

I have been having a rash of problems when I do R or RStudio updates. I get a lot of these problems over the next few sessions after an update. Is the problem apparent on this information? If not, how can I diagnose it?

Problems take the following forms:

1. Packages that are present in my R library give errors as missing. Most often these are tidyverse packages.
2. Certain dependencies of tidyverse are not installed, especially stringi . Similar problems occur less often with Hmisc .
3. Packages are correctly downloaded, but I get an error stating that the older version could not be removed. These errors vanish if I remove the packages indicated by hand and install them individually.

4. I get similar problems if I try to update tidyverse with tidyverse_update , eg:

   tidyverse::tidyverse_update(recursive = TRUE) Error in loadNamespace(j <- i[[1L]], c(lib.loc, .libPaths()), versionCheck = vI[[j]]): there is no package called 'readr'

This although the readr package is in fact present.

5. I just now got a new error message on trying to install readr individually:

   Warning in install.packages: lzma decoding result 10 Warning in install.packages: unable to access index for repository https://cran.case.edu/bin/windows/contrib/3.5 : Line starting 'Pac ...' is malformed!

And on trying again:

Error in install.packages : invalid version specification ‘NA’

I usually start a session with update.packages(ask=FALSE) . I have RStudio set up to always run as Administrator. If asked, I generally install from binaries rather than source.

I install and load groups of packages using the following functions:

# Install packages (if not already installed)
in_packs <- function(packs){
  new_packs <- packs[!(packs %in% installed.packages()[,"Package"])]
  if(length(new_packs) < 1) print("Already installed.") else {
    install.packages(new_packs, dependencies = TRUE,      # move to fairR_fns 
                        repos=c("https://cran.case.edu/",
                                "https://cran.rstudio.com"))

# Load packages (if not already installed)
load_pks <- function(pks){
  invisible(sapply(X = pks[seq_along(pks)], FUN = library, character.only = TRUE))
}

Here is my session information.

R version 3.5.1 (2018-07-02)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.0       lubridate_1.7.4  lattice_0.20-38  tidyr_0.8.2     
 [5] assertthat_0.2.0 digest_0.6.18    R6_2.3.0         plyr_1.8.4      
 [9] backports_1.1.3  httr_1.4.0       ggplot2_3.1.0    pillar_1.3.1    
[13] rlang_0.3.0.1    lazyeval_0.2.1   stringr_1.3.1    munsell_0.5.0   
[17] broom_0.5.1      compiler_3.5.1   modelr_0.1.2     pkgconfig_2.0.2 
[21] tidyselect_0.2.5 tibble_1.4.2     crayon_1.3.4     dplyr_0.7.8     
[25] grid_3.5.1       nlme_3.1-137     jsonlite_1.6     gtable_0.2.0    
[29] magrittr_1.5     scales_1.0.0     cli_1.0.1        stringi_1.2.4   
[33] bindrcpp_0.2.2   generics_0.0.2   tools_3.5.1      forcats_0.3.0   
[37] glue_1.3.0       purrr_0.2.5      hms_0.4.2        yaml_2.2.0      
[41] colorspace_1.3-2 memoise_1.1.0    bindr_0.1.1      haven_2.0.0     

Answer 1

I have found that most, though by no means all, of these problems go away if I do the following:

1. Make sure I have only one instance of R open.
2. Set up both R and RStudio to run as administrator.

This is a little tricky because the instantiation of RI am running is usually opened by RStudio, not by me. For both RStudio and R, right-click on the shortcut, either on the desktop or the start menu. On the shortcut tab, select advanced. Then check the box for run as administrator. I also allow the R directory in program files to be modified by other programs, though I am not sure this was necessary. Right-click, properties, Security tab, select ALL APPLICATION PACKAGES, click on edit, check on Full control.

3. Set the R folder itself, and all its sub-folders, to shared. (Be aware that this creates a security risk if you are on a shared machine). Right-click on the R directory -- usually the one in your "My Documents" folder, not the one in Program Files -- and under the sharing tab, hit share.

Of the problems that remain when these things have been done, most seem to be caused by, usually short-term, problems with the repository copy. Such problems can disturb not only the corrupted (or whatever) package, but also its dependencies. Delete the folder of the problem package manually, restart R and, before loading any packages, switch to a different repository. Then install as usual