“R is a rich statistical environment, released as free software, which includes a programming language, an interactive shell, and extensive graphing capabilities. This installment discusses creating reusable and modular components for R development. This article follows up two prior installments and looks at the object-orientation in R along with some additional general programming concepts in R.”
I really like R. I have used it quite a bit. I think it’s a fantastic piece of open source software. What tends to bug me though is that few people realize that R is a clone of the open source statistical package S-Plus. S-Plus is one of the few industry-standard statistics packages available (along with minitab and SAS). R is to S-Plus as Octave is to Matlab. Often articles tend to describe the features of R as ‘inovative’ or ‘revolutionary’ when in fact said features were first to be used in in S-Plus and then ported to R. Now the articles may not use that wording, but that’s the general idea that’s gotten across.
What is interesting though is that R is taken such a strong interest in the statistics field. So much so that when companies sell additional packages for S-Plus, they are sold as ‘S-Plus/R packages’. I applaude the R team for being able to produce such an outstanding and incredibly compatible product. You don’t really see this with Octave and Matlab. Octave is often seen as a ‘really cool toy’ that engineers play with but for the real work they use Matlab. R, on the other hand, is used just as much if not more than S-Plus (I would think more). Good job on a quality product R folks.
The underlying theme here though is an epidemic I think that tends to rear it’s ugly head within open source software. And that is ‘Hey, this thing we pay money for is cool, so let’s do the same thing for free’. I think open source software won’t really be able to overcome this until the genuine inovation comes from open source in the first place.
I suppose I’m trying to say alot of things here but can’t really pinpoint any one idea in particular. Sure it may be off topic, but there you have it. My views on both R and open source in general.
I always thought that the history of R is a bit different from what you describe. Now, I can’t find any sources, so this is completely unreliable. The story as I know it was told to me by my R crouse instructor. I thought S used to be free (although maybe not Free), and many open-source type developers were contributing to S. Then, the owners of S decided to close it up, and thus the community was quite disgruntled. That’s why they developed R. You are right when you say that R’s development is typical of OS development, but not in the way you say. This is often what happens: there’s a company that creates a product (XFree, Bitkeeper, S, etc) that is freely available to the public. Then, the company decides it could make more money if instead of providing a product for free, it could charge (or make other restrictions to the license), and so the OS community decides to make their own (Xorg, R, etc). This is also why everybody used to think Qt was such a bad idea. Now, these issues have been resolved for Qt (asbestos sentence, prevents flames). In other words, I disagree entirely with what you’re saying.
Also, I think the OS-correlate to matlab is scilab, and I often hear of people that actually use scilab as a viable alternative.
Finally, I think you left out one of the biggest players in statistics: SPSS. So when you say there are only few industry standard software packages, I think you are wrong. We already mentioned 3-and-a-half (including R/Splus). That’s way more than office software packages, for instance.
I stand almost entirely corrected. I was basing too off of information told to me by various figures in my stat dept. How wrong they and I were. After research on my own, I found that in fact, R is just another implementation of the S language. The S language was a language developed for statistics and upon it’s initial conception provided an interpreter (at the time also just referred to as S). From there, two other primary interpreters were developed. S-Plus as a commercial package and R as an open package.
As it was based on the definition for the S language, It was neither innovative nor a clone of a commercial package. So R is infact a terrible example of the point I was trying to make. I appologise for running my mouth (or fingers as it were) without doing my own research and verification of the facts.
Bah, the articles linked described quite clearly R’s origins so this is a counter example of what you’re saying.
As for open source being only a reinvention of closed source and not innovative, given that Internet is based on open source programs frankly this is quite a ridiculous statement, but I suppose that there is no better blind that the one who doesn’t want to see..
Plus frankly “innovation” is overrated most open or closed source are better/different execution of old concepts, as long as the result is better than the original who cares if it isn’t truly innovative?
For me, R developpers clone S for the same reason that OOo “clone” word: for compatibility with installed base, not because S or Office are particulary innovative.
Also, I think the OS-correlate to matlab is scilab, and I often hear of people that actually use scilab as a viable alternative.
eh. sci lab is sort of a viable alternative but its more of a in-the-style-of matlab, versus octave which is pretty much a reimplimentation like R is to S