Comments on: My R naming nemesis

By: grubert

grubert — Thu, 27 Dec 2012 20:10:59 +0000

It annoys me when a language designer decides to break with decades of tradition and employ a character for a radically different purpose.

It smacks of arrogance, crankiness, and a lack of genuinely original thought.

By: Derek Jones

Derek Jones — Thu, 20 Dec 2012 11:30:44 +0000

@Jens Oehlschlägel That is certainly a useful naming convention and further adds to the feeling that dot is an operator. However, there is nothing to stop developers using this character in other contexts and I have seen some who seem to use this character where others would use underscore.

By: Jens Oehlschlägel

Jens Oehlschlägel — Wed, 19 Dec 2012 16:12:47 +0000

It seems to me that what misses in this discussion is the function of dots in R naming: dots in R function names serve as a means to realize R’s oldest class system (S3 classes). A METHODNAME follows the pattern GENERICNAME.CLASSNAME as in summary.lm, where calling GENERICNAME(OBJECT OF CLASSNAME) will dispatch to GENERICNAME.CLASSNAME if present and to GENERICNAME.default otherwise. This allows for something like overloading in other languages.

(resubmitted with placeholders in uppercase since angle brackets dispear here)

By: Dinre

Dinre — Tue, 18 Dec 2012 15:33:01 +0000

As an engineer myself, I personally don’t have a problem with the ‘.’ in names, but I agree that it can be jarring at first. However, I think the real issue is to be obviously consistent with whatever naming convention you use. For instance, I have a problem with ‘foo’ being an object by itself with ‘foo.bar’ being a different and completely unrelated object. I see this a lot in people’s R code, and it makes reading very difficult for me. Some popular programming languages use the ‘.’ to denote membership where ‘foo.bar’ indicates [bar] as a member of object/class [foo], so I avoid doing this in my own code.

I’ve read a number of books on R, and everyone seems to use a slightly different naming convention. The difference between the books I liked and the ones I struggled with had mostly to do with how obvious the naming conventions were. The more obvious and consistent the naming convention, the easier the code was for me to read.

By: Robert Young

Robert Young — Tue, 18 Dec 2012 03:54:32 +0000

Now, how do we get used to the mix of noun.verb and verb.noun naming?