How to determine which older version of the R package is compatible with my R version

Question

I am trying to install the "tm" package but then I get an error saying that "tm" is not available for my R version

package ‘tm’ is not available (for R version 3.0.2) 

But then I saw that someone suggested I download the archived version from

http://cran.r-project.org/src/contrib/Archive/tm/?C=M;O=A 

and then try installing from source.

My question is how do I determine which file there in the list is compatible with my R version?

Answer 1

I developed an answer related to the approach here, but which uses only base R (you don't need XML or devtools or anything). It also potentially handles some contingencies that may not be addressed by the solution I linked to from the other question. Since it was around 100 lines of code, rather than just post a huge function here, I rolled it into a package oldr you can get from GitHub here:

oldr package GitHub repo

The package has just one exported function, install.compatible.packages(). I have tested it on Ubuntu 18.04. I installed R 3.1.0 and installed tm (and its dependencies). The current version of tm requires R 3.2.0, so wouldn't be available via install.packages(), but my function allows its installation:

> oldr::install.compatible.packages("NLP") Installing package into ‘/home/duckmayr/R/x86_64-unknown-linux-gnu-library/3.1’ (as ‘lib’ is unspecified) * installing *source* package ‘NLP’ ... ** package ‘NLP’ successfully unpacked and MD5 sums checked ** R ** inst ** preparing package for lazy loading ** help *** installing help indices ** building package indices ** testing if installed package can be loaded * DONE (NLP) > oldr::install.compatible.packages("slam") Installing package into ‘/home/duckmayr/R/x86_64-unknown-linux-gnu-library/3.1’ (as ‘lib’ is unspecified) * installing *source* package ‘slam’ ... ** package ‘slam’ successfully unpacked and MD5 sums checked ** libs gcc -I/opt/R/3.1.0/lib/R/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c apply.c -o apply.o gcc -I/opt/R/3.1.0/lib/R/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c grouped.c -o grouped.o gcc -I/opt/R/3.1.0/lib/R/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c sparse.c -o sparse.o gcc -I/opt/R/3.1.0/lib/R/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c util.c -o util.o gcc -shared -L/usr/local/lib -o slam.so apply.o grouped.o sparse.o util.o -L/opt/R/3.1.0/lib/R/lib -lRblas -lgfortran -lm -lquadmath -L/opt/R/3.1.0/lib/R/lib -lR installing to /home/duckmayr/R/x86_64-unknown-linux-gnu-library/3.1/slam/libs ** R ** preparing package for lazy loading ** help *** installing help indices ** building package indices ** testing if installed package can be loaded * DONE (slam) > oldr::install.compatible.packages("tm") Installing package into ‘/home/duckmayr/R/x86_64-unknown-linux-gnu-library/3.1’ (as ‘lib’ is unspecified) * installing *source* package ‘tm’ ... ** package ‘tm’ successfully unpacked and MD5 sums checked ** libs gcc -I/opt/R/3.1.0/lib/R/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c copy.c -o copy.o gcc -shared -L/usr/local/lib -o tm.so copy.o -L/opt/R/3.1.0/lib/R/lib -lR installing to /home/duckmayr/R/x86_64-unknown-linux-gnu-library/3.1/tm/libs ** R ** data ** inst ** preparing package for lazy loading ** help *** installing help indices ** building package indices ** installing vignettes ** testing if installed package can be loaded * DONE (tm) 

Update: Testing on Windows

I now have also had the chance to test on Windows (8.1), and everything worked smoothly for installing tm to an old version of R (v. 3.1.0):

> oldr::install.compatible.packages("NLP") Installing package into ‘C:/Users/User/Documents/R/win-library/3.1’ (as ‘lib’ is unspecified) trying URL 'http://mirror.las.iastate.edu/CRAN/bin/windows/contrib/3.1/NLP_0.1-9.zip' Content type 'application/zip' length 278699 bytes (272 Kb) opened URL downloaded 272 Kb  package ‘NLP’ successfully unpacked and MD5 sums checked  The downloaded binary packages are in         C:\Users\User\AppData\Local\Temp\RtmpojDNlF\downloaded_packages > oldr::install.compatible.packages("slam") Installing package into ‘C:/Users/User/Documents/R/win-library/3.1’ (as ‘lib’ is unspecified) trying URL 'http://mirror.las.iastate.edu/CRAN/bin/windows/contrib/3.1/slam_0.1-32.zip' Content type 'application/zip' length 111528 bytes (108 Kb) opened URL downloaded 108 Kb  package ‘slam’ successfully unpacked and MD5 sums checked  The downloaded binary packages are in         C:\Users\User\AppData\Local\Temp\RtmpojDNlF\downloaded_packages > oldr::install.compatible.packages("tm") Installing package into ‘C:/Users/User/Documents/R/win-library/3.1’ (as ‘lib’ is unspecified) trying URL 'http://mirror.las.iastate.edu/CRAN/bin/windows/contrib/3.1/tm_0.6-2.zip' Content type 'application/zip' length 710798 bytes (694 Kb) opened URL downloaded 694 Kb  package ‘tm’ successfully unpacked and MD5 sums checked  The downloaded binary packages are in         C:\Users\User\AppData\Local\Temp\RtmpojDNlF\downloaded_packages 

Update: Additional Parameters

Now users can specify which version of R to attempt installation for (R_version parameter), and which directory to install packages to (lib parameter), which could be useful for testing or other purposes.

Answer 2

You can use the METACRAN mirror:

Go to the blame page of the DESCRIPTION file of the package you're interested in.

E.g. for tm: https://github.com/cran/tm/blame/master/DESCRIPTION

Find the Depends line and click as many times as needed on the View blame prior to this change icon, until an old enough R version is displayed.

If you want to automate that, it may be better to use crandb (also from METACRAN).

Side note: sometimes package authors list R (≥ x.y.z) as a dependency just to be safe because they use version x.y.z and didn't do any tests with previous versions.