Re: Problem with updating packages - SOLVED

Hello,

Inline.

Às 10:36 de 28/05/2026, Karl Schilling escreveu:
> Thank you everybody so much for your help.
> 
> I adjusted my CRAN mirror in my etc/Profile.site file as suggested by 
> Rui Barradas - that solved the issue with the twice repeated question of 
> which repository to use. 

I'm glad it was helpful.

Whenever I have problems with update.packages() the following script 
usually works. No guarantees, though.


inst_pack <- function(pkgs, lib = .libPaths()[1],
                       repos = "https://cloud.r-project.org",
                       Ncpus = 1L){
   utils::install.packages(
     pkgs = pkgs,
     lib = lib,
     repos = repos,
     Ncpus = Ncpus
   )
}

repo_default <- "https://cran.r-project.org"
repo_cloud <- "https://cloud.r-project.org"

## set repo, use default if you prefer it
local({r <- getOption("repos")
   r["CRAN"] <- repo_cloud
   options(repos=r)
})

#
inst_pkgs <- installed.packages(priority = NA_character_)
inst_pkgname <- inst_pkgs[, "Package"]
inst_version <- inst_pkgs[, "Version"]
#
avail_pkgs <- available.packages(repos = repo_cloud)
avail_pkgname <- avail_pkgs[, "Package"]
avail_version <- avail_pkgs[, "Version"]

i <- which(avail_pkgname %in% inst_pkgname)
j <- which(inst_pkgname %in% avail_pkgname)

any(avail_pkgname[i] != inst_pkgname[j])
# FALSE

# coerce the version strings to real version numbers
avail_version <- package_version(avail_version)
inst_version <- package_version(inst_version)

# which packages to update
i_update <- which(avail_version[i] > inst_version[j])
length(i_update)

ncores <- 1L
# ncores <- max(1L, parallel::detectCores() - 4L)
to_update <- avail_pkgname[i][i_update]
inst_pack(to_update, Ncpus = ncores)



Hope this helps,

Rui Barradas

It did not resolve, however, the issue that all
> packages were flagged as  not actual.
> 
> So I followed the advice of Jeff Newmiller - actually I went one step 
> further and completely de-installed my R version, re-installed it, and 
> then downloaded the packages I need. That solved the problem.
> 
> 
> One point that I still do not understand is why AFTER updating my 
> original R 4.5 packages to 4.6, those were still identified as not up-to 
> date. So why I was in this "ndless" loop of ineffective (?) updating. 
> But maybe I should be pragmatic and be satisfied that it now works again.
> 
> Again, thank you all for your help.
> 
> Best,
> 
> Karl Schillng
> 
> On 28.05.2026 00:33, Uwe Ligges wrote:
>>
>>
>> On 28.05.2026 00:02, Jeff Newmiller wrote:
>>> So can we interpret this to mean that update.packages() will fix 
>>> (replace) old packages that were copied into the new library? What if 
>>> one or more packages were dropped/archived or pulled from non- 
>>> standard package sources?
>>
>> No, not if a package is archived from CRAN or downloaded from a 
>> repository that is not declared. update.packages() won't find that the 
>> update is needed in such a case
>> I talked about active CRAN packages.
>>
>> Best,
>> Uwe Ligges
>>
>>
>>>
>>>
>>> On May 27, 2026 2:06:12 PM PDT, Uwe Ligges <[email protected] 
>>> dortmund.de> wrote:
>>>
>>>
>>>
>>>     On 27.05.2026 18:01, Karl Schilling wrote:
>>>
>>>         I am running R under Windows 11. recently updated R to the
>>>         4.6patchhed version. I then copied my packages from my previous
>>>         version (4.5) to the library of 4.6. Then I updated all 
>>> packages.
>>>
>>>         Since then, I see the following behavior:
>>>
>>>         Each time I run
>>>         "update.packages(ask='graphics',checkBuilt=TRUE)" I am asked 
>>> TWICE
>>>
>>>         "--- Please select a CRAN mirror for use in this session ---".
>>>
>>>         And, more embarrassingly, all my Cran and Bioconductor are said
>>>         to require an update. That also happens when I run
>>>         "update.packages", say, one hour after my last update.
>>>
>>>         And when I run "old.packages()", it seems that all my packages
>>>         are identified as being old.
>>>
>>>         Any suggestions what might be going on?
>>>
>>>
>>>     For a major version change (as from 4.5.x to 4.6.y) we do not
>>>     guarantee API compatibility, hence we cannot guarantee that a
>>>     package built for 4.5.x will work under 4.6.y. As a consequence,
>>>     update.packages() knows that everything has to be reinstalled.
>>>
>>>     Best,
>>>     Uwe Ligges
>>>
>>>
>>>
>>>
>>>
>>>
>>>         Thank you so much in advance,
>>>
>>>         Karl Schilling
>>> ------------------------------------------------------------------------
>>>         [email protected] mailing list -- To UNSUBSCRIBE and more, 
>>> see
>>>         https://stat.ethz.ch/mailman/listinfo/r-help <https://
>>>         stat.ethz.ch/mailman/listinfo/r-help>
>>>         PLEASE do read the posting guide https://www.R-project.org/
>>>         posting- <https://www.R-project.org/posting-> guide.html
>>>         and provide commented, minimal, self-contained, reproducible 
>>> code.
>>>
>>>
>>> -- 
>>> Sent from my phone. Please excuse my brevity.
>>