Packages tested with pqR

Radford Neal edited this page Jul 2, 2017 · 38 revisions

This page reports on R packages that do and do not work with pqR, and describes a repository of packages known to work.

A repository of packages that work with pqR

A collection of packages that are known to work (at least on a Linux system) with the current stable release of pqR is available from the repository. The base and recommended packages distributed with pqR are also known to work (they are not included in the repository). Repositories for some specific versions of pqR can be obtained by appending the release date, as /YYYY-MM-DD, to the URL.

These packages pass all checks (including running vignettes) on an Ubuntu 12.04 system with a 64-bit Intel processor, using uncompressed pointers, except that a small number of packages can't complete some checks because they need other packages that are not in this repository, and a small number of checks fail for reasons known to be innocuous. Also, some packages fail unless run in a UTF-8 or Latin-1 locale (not C/POSIX), some packages require software that is not installed by default with Ubuntu 12.04, and some packages may fail if an external URL cannot be found. Finally, some packages need to be installed with R_PARSE_DOTDOT set to FALSE (see help("..") for details).

The repository currently contains 6365 contributed packages, which are in addition to the 14 base packages and 16 recommended packages distributed with pqR. The names and versions of the contributed packages are listed here.

To use the repository when installing packages, simply include the argument repos="" in the call of install.packages. For example:

install.packages ("ggplot2", repos="")

Packages that do not work with pqR

A few packages are known to fail with pqR-2017-06-09 because of incompatibilties of pqR with R-2.15.1. A few more packages do dubious things that don't work in pqR, but that happen to work in R-2.15.1, though it is clear that such things are not intended to work even in R-2.15.1.

Of course, a number of packages state that they depend on a version of R later than 2.15.1, which is the latest version that pqR aims to be compatible with. These have not been tested with pqR (though some might work, if their declared dependence is actually unnecessary).

Below, are some packages that do not work with pqR. Of course, packages that depend on these packages will also not work with pqR.

Also, Rcpp does not work with pqR if compressed pointers are enabled.

data.table - current version 1.9.6 and earlier versions.

Version 1.8.10 and earlier versions of this package contained code that modified an object without first checking that it was unshared (ie, had NAMED of 0). This happened to work in R-2.15.0, but not in pqR, and probably not in R-3.1.0. Version 1.9.6 has eliminated some such object modifications, but apparently not all, since it also doesn't pass its tests with pqR-2016-10-05.


Fails in pqR unless options(keep.parens=TRUE) is used because it relies on parentheses being preserved in expressions such as x %>>% (a ~ b), but by default pqR now deletes such parentheses (when they are needed to override precedence, and hence will be automatically added if the expression is deparsed).

For some packages with small, easily-fixable incompattibilities, the pqR repository has a new version with the necessary fixes, which do not alter functionality. The version number is altered to recognize the changes. These packages are the following:

LIM, R.huge, devtools, jpeg, likeLTD, memisc, phylosim, rJava, robustfa, testthat

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.