[Rd] Testing development code?

[Prof Brian Ripley]

On Fri, 6 Jun 2008, Spencer Graves wrote:

> Hi, R Core Team:
>     How do you test the latest changes to the core R code?  Is it just "R 
> CMD check"?  Or do you use other tools, like the perl "prove" module?
>     I ask, because I'm about to start developing routine testing for other 
> (non-R) software, for which I can't do "R CMD check", and I'd like to know 
> what you do.

'make check-devel', 'make check-all' and regression tests of (essentially) 
R CMD check over CRAN/BioC (the latter in arrears via daily checking 
unless the change is thought likely to break things).

If we were starting from scatch we would probably use a unit-testing 
framework such as provided by Runits.

C-based projects often use DejaGnu and similar -- scripting languages 
usually use themselves to run their test suites.


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595