RLint: Reformatting R Code to Follow the Google Style Guide
Tuesday, 08 July 2014
Google Research has shown off software that will check and reformat R code to conform to the Google R Style Guide.
The project was discussed in a paper given by Andy Chen of Google Research at this year’s R User Conference in Los Angeles.
R is rapidly becoming the top language for statistics and data processing in general and it needs to be taken seriously as it is being used to create large systems often by users who aren't primarily programmers. Other languages have "lint" programs designed to remove the irritating "fluff" that obscures meaning and degrades the code - so why not R?
Chen's talk showed how RLint has been developed by Google to check and reformat R code, warning of violations and optionally changing the layout to produce compliant code.
Code elements RLint looks at include proper spacing, line alignment inside brackets, and other style violations. Chen says RLint is automatically used within Google as part of the peer review process for R code.
The need for consistent coding layout and format within Google is clear when you consider the statistics Chen quotes, saying that within Google around 40 percent of R files are modified by more than one programmer.
Google has an R Style Guide that specifies best practices for naming identifiers; program layout in terms of indentation and spacing; how to comment a program and how to comment functions. The measure of success, according to the guide, is that any programmer should be able to instantly understand structure of any code.
RLint was developed in Python with the goal of minimizing the overhead of following the style guide. When used within Google, all code violations of the style guide are flagged by the code review tool, and violations must be corrected before code submission.
Chen says Google encourages CRAN package authors and other R programmers to use this tool. It is driven from the command line and runs in Linux, Unix, Mac or Windows.
Speculating on what might have happened if Babbage had built his Analytical Engine is fun, but did you ever think that a Victorian computer could implement a neural network and learn to read handwritt [ ... ]
Version 1.7 of the Perl 5 to C/C++ optimizing compiler, codenamed Tycho was released earlier this month. The issue RPerl tries to address is Perl's slow performance, sufficient for most cases, bu [ ... ]