Point well taken --- grid::gpar() is also a good example; I'll make use of your suggestion in my future coding.
Best, baptiste On 27 February 2010 15:02, Barry Rowlingson <b.rowling...@lancaster.ac.uk> wrote: > On Sat, Feb 27, 2010 at 11:29 AM, Gabor Grothendieck > <ggrothendi...@gmail.com> wrote: >> Or use modifyList which is in the core of R. > > All these solutions appear to be adding on more and more code with > less and less semantics. Result: messy code which is harder to read > and debug. > > It seems that the arguments should have proper constructors with > meaningful names. A bit like an optimisation function with a control > argument. What's better: > > o = optimmy(f, control=list(niter=10,ftol=0.01)) > > or > > o = optimmy(f,control=control(niter=10,ftol=0.01)) > > here you are explicitly constructing a 'control' object that has the > options for controlling the optimiser, and it will have its own > sensible defaults which the user can selectively override. It seems to > be the correct paradigm for collecting related parameters, and indeed > is used in lots of R code. > > Done this way you get selectively overridable arguments, a meaningful > name, readable code, leveraging the built-in defaults system, and the > possibility of sticking consistency checks in your control() function. > > Tell me where that is not made of pure win? > > Barry > ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
