Question

I noticed that this answer to a question about .Rprofile settings mentions setting two repositories for CRAN.

Is there any benefit to doing so, such as the possibility of one site being down, then another is accessed? It seems that the CRAN mirrors are identical or eventually mirrors, and uptime seems very good, so are there any other advantages? (I realize OmegaHat and Bioconductor have their own repositories, so I'm focusing on CRAN.)

Is it fair to say that CRAN is for stable releases and R-Forge is a staging area? If so, then it seems that including R-Forge as a repository is more suitable for instances of beta testing various packages.


Update 1: This post has some good links on comparing mirrors, especially determining if two or more mirrors are out of sync (i.e. outdated). This is the direct link for information on the availability status of mirrors, though it doesn't reflect the status of packages on the mirrors. Dirk's answer is useful for determining discrepancies in the packages offered.

Was it helpful?

Solution

A) I don't see any particular advantage to having multiple repositories. If you choose a reliable one and it is behind the times or temporarily out of service. it is pretty much trivial to access another. I generally use the CMU repository, but the Berkeley repo also seems reliable.

B) Agree that CRAN (or mirrors) is for stable releases and r-forge is a "staging area". If you need a more recent version or one that is not available, then r-forge is the first location to consult. The bioconductor repos may also be accessed.

It is a good idea to know how to access the CRAN package check page.

OTHER TIPS

If fault tolerance and flexibility is important, it can be useful to set local, internal, and remote repos.

repos["CRAN"] <- c("file://opt/cran","https://cran.example.com","https://cran.r-project.org")
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top