To improve performance of difference operations, and especially of stability scores, pkgdiff uses a cache of pre-processed package information on Github. This cache greatly speeds up the operation of pkgdiff functions. The cache is created by a periodic process that extracts information from packages on CRAN, and stores it in a way that the information can be quickly retrieved when a pkgdiff function is called.
Note the following about the pkgdiff cache:
To access the cache, the computer pkgdiff is running on must have access to the internet and, specifically, Github. If your machine is isolated from the internet, or blocked from Github, pkgdiff will error.
The cache does not contain all CRAN packages. It contains packages that exceed 1000 downloads per month. These packages constitute approximately 20% of the top CRAN downloads, or about 4000 packages. For most R users, any package you might be interested in will be included in this subset.
To find out which packages are in the pkgdiff
cache, you may call the function pkg_cache(). See the
documentation of that function for additional information.
If a package is not in the cache, all pkgdiff functions will still work. However, they will be much slower than packages that are included in the cache.
You may request that a package be added to the cache by submitting an issue to the pkgdiff issue log here.
Observe that the pkg_info() function also returns a
parameter on whether or not the selected package is in the cache. It is
recommended to run pkg_info() or pkg_cache()
first, and check if the package is in the cache before running
pkg_stability().
If a package is not in the cache, you may speed up
pkg_stability() by setting either the “releases” or
“months” parameters. Either of these parameters will limit the scope of
the call to pkg_stability(). For example, if a package has
a lifespan of 10 years, you can greatly reduce the amount of stability
processing by reducing the comparison period to 36 months instead of the
entire 10 years. Do note, however, that such scope reductions may also
have an effect on the calculated score and assessment.
The package cache is updated periodically, usually every day. The update looks for new versions of packages that are already in the cache. If a new version is found, the update routine will add the new version to the stored infos for that package, and also update the stability data. The update routine also looks for new packages that have recently met the 1000 downloads per month threshold. If such a package is found, it will be added to the cache automatically. Packages that fall below the 1000 downloads per month threshold are not removed. Once a package has been added to the cache, it will be retained indefinitely.