Skip to content

Problem with compatible library versions

7 messages · Sabine Braun, Uwe Ligges, Richard O'Keefe +1 more

Original
Sabine Braun
# 
On the github website I have reported several bugs with new versions of 
the tidyverse group (probably dplyr) which prevent me from using R 
normally. I wanted to go back to older versions but this seems not bo be 
easy. I downloaded R 4.1.2. and Rtools 40 but the library versions 
installed are still the newest ones. I was able to install dplyr 1.0.7. 
manually but there are error messages on incompatibility when loading 
this version. Is there a possibility to load older library versions 
which alre compatible ?

Thank you very much!

Best regards

Sabine Braun
Uwe Ligges
#
On 10.10.2023 17:34, Sabine Braun wrote:
So this is on Windows.

Actually, if you install R-4.1.2 and use a clean library and install 
binaries, then you should get binary installation from CRAN that fit to 
the R-4.1.x series.

If you want to install older package versions, then you have to install 
these one by one from sources, unfortunately.

Best,
Uwe Ligges
Richard O'Keefe
# 
There is a fairly straightforward way to load older versions
of packages, and that is to use the 'groundhog' package.
As the first sentence of https://groundhogr.com/ puts it:
   Make your R scripts reproducible by replacing library(pkg)
   with groundhog.library(pkg, date).

pkg can be a vector of package names or a single name.

On Wed, 11 Oct 2023 at 20:58, Uwe Ligges <ligges at statistik.tu-dortmund.de>
wrote:
http://www.R-project.org/posting-guide.html

  
  


  


      

    

    

      
      

        


  

        

      Ebert,Timothy Aaron
    

    

      
      #
    

  


  

      
Is that a method where a program that I write today would still run without changes in 10 years?
Tim

-----Original Message-----
From: R-help <r-help-bounces at r-project.org> On Behalf Of Richard O'Keefe
Sent: Wednesday, October 11, 2023 8:08 AM
To: Uwe Ligges <ligges at statistik.tu-dortmund.de>
Cc: r-help at r-project.org
Subject: Re: [R] Problem with compatible library versions

[External Email]

There is a fairly straightforward way to load older versions of packages, and that is to use the 'groundhog' package.
As the first sentence of https://groundhogr.com/ puts it:
   Make your R scripts reproducible by replacing library(pkg)
   with groundhog.library(pkg, date).

pkg can be a vector of package names or a single name.

On Wed, 11 Oct 2023 at 20:58, Uwe Ligges <ligges at statistik.tu-dortmund.de>
wrote:

            

      
      
      
    

        

      
      
    

        
http://www.r-project.org/posting-guide.html

        

      
      
    

        
______________________________________________
R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.r-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

  
  


  


      

    

    

      
      

        


  

        

      Uwe Ligges
    

    

      
      #
    

  


  

      
No.

In 10 years you may have different hardware and a different OS, so that 
the old R and old packages won't run on the new hardware or produce 
different results. This may even happen for some sort of containers and 
virtualizations.

To be really safe, you ideally need to keep the whole system as is (but 
then there are security updates etc).

Best,
Uwe Ligges

        
On 11.10.2023 14:54, Ebert,Timothy Aaron wrote:

            

      
      
      
    

        

      
      
    

        

      
      
    

  
  


  


      

    

    
1 day later

    

      
      

        


  

        

      Richard O'Keefe
    

    

      
      #
    

  


  

      
The purpose of 'groundhog' is to support reproducible research.
As Uwe Ligges pointed out, there are practical limits to that.
Since I am typing this on a 10-year-old machine, it's clearly
feasible in *some* instances, but as an illustration of Uwe's
point, Ubuntu still supports 32-bit applications, but it does
not support 32-bit x86 hardware (since 20.04 IIRC).

I think it is fair to say that 'groundhog' *supports*
reproducible research, but nothing can *guarantee* it.

        
On Thu, 12 Oct 2023 at 01:54, Ebert,Timothy Aaron <tebert at ufl.edu> wrote:

        

            

      
      
      
    

        

      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Sabine Braun
    

    

      
      #
    

  


  

      
Thank you very much :-)), this worked! Now the loaded libraries are 
compatible with each other.

Best regards

Sabine Braun

Am 11.10.2023 um 14:08 schrieb Richard O'Keefe: