[R-pkg-devel] Looking for info on how CRAN package developers handle breaking changes with other packages

[Chris Bogart]

Hi, I'm looking for some help from developers who write R packages. My
research group is interested in the choices package managers and
repositories are make when designing their ecosystems. Newer packaging
and modules systems like Node.js/NPM, Google's go language, Python's
PyPI, are making very different design choices from each other, and
they are very different (and probably informed by) older packaging
systems like R's CRAN and Perl's CPAN. We'd like to know what the
impact of those design choices are on how you negotiate breaking
changes among R package developers when the packages depend on each
other. We're also interested in what values you think CRAN's design is
trying to optimize.

Could I ask people to take ~20 minutes of your time and fill out our
survey at http://breakingapis.org/survey? We'll report back to the R
community the results (or there's a sign up link to be notified when
results are out at http://breakingapis.org).

(If you distribute R packages in other ways, like Bioconductor or
devtools/Github; or if you use other ecosystems altogether such as
Node.js/NPM, Python/PyPI, Hackage, or something else, we're still
interested -- just make a different "software ecosystem" choice on the
first page. The study is a comparison among ecosystems; CRAN is one
point of comparison).

Thank you,
Chris Bogart
Institute for Software Research
Carnegie Mellon University
cbogart at cs.cmu.edu