Talk:R (programming language)/Archive 1

Advert tag
Even Free Software articles suffer from advertitis on here. I'll try to tidy this later. Chris Cunningham 00:43, 3 August 2006 (UTC)


 * You added the comment, after a line discussing R's interface, "what are we discussing here? The compiler? The IDE? This line can probably be removed if it's just the compiler". Although there are a couple of experimental compilers, R is typically run as an interpreted language in an interactive session. Submit a line of code (or maybe a few lines), read the output, submit some more, and so on. Some of the usual "programming language" connotations don't really apply here, which is one reason why it might be good to change the article's title (see above). -- Avenue 11:20, 3 August 2006 (UTC)
 * Right - in that case a description of the normal runtime usage would be a great thing to elaborate on. Chris Cunningham 12:01, 3 August 2006 (UTC)

Quick tutorial
Why was it eliminated? I had a hard time finding how to start using R. I still do - it seems that every help file or documentation is directed either to those that know nothing or to those that know everything. Albmont 01:35, 30 September 2006 (UTC)


 * While your description of R documentation is astute (unfortunately), WP essentially is not a venue for "how to" information. It is a venue for "what is" information.  Note I didn't delete it (nor would I), but if called on a user vote, would agree to delete for that reason.  Baccyak4H 02:41, 30 September 2006 (UTC)

I had learned SAS programming in school, but cannot afford to purchase it, so was very happy to find R,..., until I tried to learn how to use it. There is a real need to have a bare bones, easy introduction to R. At present, what I find as free information on how to use it drops me in the middle of a thick forest. If I could vote, I would say put back that Quick tutorial.
 * I'm in the same situation, and have found R a difficult study as well. Purchasing the Crawley (2005) book cited in the main article was something of a breakthrough for me.  A set of exercises for the book are freely available on the author's website, and have some utility even unaccompanied by the book.
 * --Belgrano 14:48, 8 February 2007 (UTC)

Tutorials are not exactly encyclopedic, and should be kept off of here, but feel free to add to (the barely started) B:R Programming wikibook. The main R page ] has some good links to other tutorials and resources (much better than wiki-land has to offer). It's not the prettiest of website, but it is very resourceful. +mwtoews 17:58, 8 February 2007 (UTC)

GUI
I`ve been trying to find the GUi for R but I can`t find them. Anyone can put a link when it say that graphical user interfaces are available? where, thanks ;)--ometzit&lt;col&gt; 00:33, 13 February 2007 (UTC)


 * A few are listed in the Productivity tools section. And here's a more [comprehensive list. -- [[User:Avenue|Avenue]] 01:52, 13 February 2007 (UTC)

Mege CRAN into R (programming language)
There really isn't much to say about CRAN in a separate article. At most, it could use a small section in the R article (say R (programming language)?). +mwtoews 18:28, 6 December 2006 (UTC)
 * Done. +mwtoews 21:03, 29 December 2006 (UTC)

The article CRAN has been brought back to life. I suggest again to merge it back to this article, for the same reasons at the top of this section. + m t  03:49, 10 July 2007 (UTC)
 * I agree. -- Avenue 06:46, 10 July 2007 (UTC)
 * Done, and I double checked the four language interwiki "CRAN" articles, which all link to R (programming language) for the "en" edition. + m t  21:05, 10 July 2007 (UTC)

Robert Gentleman
The internal link (under the emblem) for Robert Gentleman directs itself to another person with the same name ( a baseball player). I am temporarily going to remove the link until a seperate article/disambiguation page is created. Grokmenow 15:24, 16 July 2007 (UTC)

Fair use rationale for Image:R gui on os x.png
Image:R gui on os x.png is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images uploaded after 4 May, 2006, and lacking such an explanation will be deleted one week after they have been uploaded, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot (talk) 20:00, 26 November 2007 (UTC)

GNU R
What is GNU R (redirect)? --Abdull (talk) 16:48, 8 February 2008 (UTC)

Merger proposal
Maybe there should be a distinction between R as a programming language and the capabilities of R in statistical analysis, but we would need to know the intentions of expansion for the "R Statistics" article. Presently the contents seems very similar. Melcombe (talk) 09:23, 13 June 2008 (UTC)


 * I think it should be ok to go ahead and merge. There isn't enough content at this point - we can always split off again later. I'll get onto this if no one else is doing it.--Fangz (talk) 15:31, 14 June 2008 (UTC)

Milestones
Why do the explanations for the Milestones stop after 2.4.1 ? It does not make much sense to simply name version numbers. --Asdirk (talk) 11:11, 8 July 2008 (UTC)
 * Hi Dirk! Done. Greetings from Vienna. -- Alfie (talk) 23:47, 8 July 2008 (UTC)
 * This section is becoming ungainly and could benefit from major summarization. For example, the repeated notes about bug fixes and "maintenance/development release" are not particularly interesting.  Even many of the technical changes (tcltk event loop with X11, etc) are not relevant for an introductory article to R. More interesting might be to note that version 2.7 was the 10th anniversary of R.  Instead of repeatedly referring to mailing list articles with the changes, why not just have one link to the NEWS file: https://svn.r-project.org/R/trunk/NEWS?  Have a look at the historical timeline section of the S-Plus page or the history section of the S page for ideas. -- kwstat.  —Preceding unsigned comment added by 170.54.58.4 (talk) 16:54, 9 July 2008 (UTC)
 * I started this section a while back to focus only on the big changes, and skipped releases that didn't offer anything of significance. I think we can cut out a few versions, as suggested above. Of course, it is sometimes a bit difficult to gauge what is important and significant from the releases. + m t  18:48, 9 July 2008 (UTC)
 * Hi kwstat! Start slicing. ;-) -- Alfie (talk) 21:49, 9 July 2008 (UTC)

Kate
I don't think Kate has a special mode for R. Refer to : http://www.google.co.nz/search?hl=en&as_q=&as_epq=R+programming+language&as_oq=&as_eq=&num=10&lr=&as_filetype=&ft=i&as_sitesearch=http%3A%2F%2Fkate-editor.org%2F&as_qdr=all&as_rights=&as_occt=any&cr=&as_nlo=&as_nhi=&safe=images

Mrfebruary (talk) 11:51, 8 July 2008 (UTC)


 * It looks like it has a R syntax highlighting module: http://kate-editor.org/syntax/2.5/r.xml. You can find it from the referenced page Calimo (talk) 16:50, 7 December 2008 (UTC)

Portal:Free software selected article: R
Just to let you know. The purpose of selecting an article is both to point readers to the article and to highlight it to potential contributors. It will remain on the portal for at least a week. The previous selected article was Sugar (GUI) - the OLPC's desktop interface which has been spun-off.

For other interesting free software articles, you can take a look at the archive of PFS's selectees. Gronky (talk) 02:16, 8 January 2009 (UTC)


 * The new selectee is Freenet - an anonymous filesharing system. Gronky (talk) 19:38, 13 April 2009 (UTC)

Larger font makes Examples section's code and graphic overlap
If you make the font significantly larger the Examples section's code and graphic overlap. -- Dougher (talk) 02:07, 29 July 2009 (UTC)

Adding a link to "R-bloggers"
Hi all,

I built a website that aggregates many blogs ( about 30 35 at the moment) who writes about R, called:

http://www.R-bloggers.com

To my knowledge, it is the only place there is to find this list of R bloggers (and get updates of their posts).

Since I built the website, I won't add the link myself. But still, is it important enough in order to be added as a link to the article about R?

Best,

Talgalili (talk) 14:35, 7 January 2010 (UTC)


 * (disclaimer: I read that blog) As an external link, this might be appropriate. I would support it more strongly if this article had a problem with many individual blogs getting added indiscriminantly, but I am not sure that problem exists.  Nonetheless, while I am hesitant in general to add such pages to articles, I would not object.  Aside, kudos for revealing your involvement and asking for community support first.  Baccyak4H (Yak!) 16:50, 7 January 2010 (UTC)
 * Thank you Baccyak for the support and the kind words! Best, Talgalili (talk) 22:27, 8 January 2010 (UTC)
 * Dear Talgalili,
 * My opinion is that there are already too much links per WP:EL, and that most of them should be removed (they could possibly appear on Wikibooks, though). Links to blogs should normally be avoided according to section 4, item 11 of WP:EL, but so is linking to multiple pages of the same website (section 1 item 4 of WP:EL), manufacturers/suppliers (4.14), open wikis (4.12), etc. Most of the current links fall under one of these criteria.
 * I tried enforcing that guideline previously (see two sections above), but I didn't get a consensus. So just ignore all rules as do the other editors of this article and many others, and go on with it. Keep in mind that someone may remove it later.
 * Calimo (talk) 10:56, 9 January 2010 (UTC)
 * PS : kudos as well for disclosing your conflict of interest. Wikipedia would certainly be better if all editors were behaving as you did.
 * Hello Calimo.
 * First, thank you for the detailed answer.
 * Regarding adding the link, I will wait another day or two to see what others are thinking about it. No point in forcing something that others would disapprove.
 * Lastly, I would like to comment about what you wrote regarding the link itself. R-bloggers, to my view, present a unique source for articles about R. It present something that no other link on the Article page offers (that is also the reason I built it in the first place).
 * I agree with you that some of the current links on the page deserve to be removed. But others, I personally feel, should go there that currently aren't. For example, the growing http://crantastic.org/ website, which is a wonderful endeavor.
 * With all the best, Talgalili (talk) 19:19, 9 January 2010 (UTC)

O.k people, I am going to add the link. Thank you for your opinion. Talgalili (talk) 22:22, 10 January 2010 (UTC)

Displaying R language in Wikipedia
I've tried to add R code on a talk page, using Wikipedia's but I get an error saying that language doesn't exist. Does anyone know to get it added? The language "S" doesn't exist either.

Quoting R code in Wiki pages would be useful, e.g. the code used to produce plots of statistical distributions like Binomial distribution. Tayste (edits) 21:38, 9 February 2010 (UTC)


 * The good news is that this is incorporated in the latest stable version of GeSHi (the syntax highlighter), v 1.0.8.6, and it seems to work okay. (For an example, go here within the next week and click on the "Highlight" button.) The bad news is that Wikipedia is still using 1.0.8.4. Here is the bug report for the last upgrade, and a new request asking for it to be upgraded again. It couldn't hurt to go vote for it. -- Avenue (talk) 02:19, 12 February 2010 (UTC)
 * Thanks for the heads up - I just went and voted... —Preceding unsigned comment added by Talgalili (talk • contribs) 05:03, 12 February 2010 (UTC)

Creating a new section (under GUI section) - "building R gui modules"
Hi all, I found there are multiple ways for building R GUI modules, such as:


 * traitr - An interface for creating GUIs modeled in part after the traits UI module for python.
 * deducer plugins
 * gWidgets - gWidgets provides a toolkit-independent API for building interactive GUIs.

And I imagine there are more.

Do you think it is worth giving them a short survey in the article?

(I wasn't sure, that's why I am asking)

Best, Talgalili (talk) 05:24, 29 April 2010 (UTC)

Software that embeds/uses R
The Feature section mentions two other software packages that use R (TIBCO Spotfire and Weka). I would suggest to remove these two and make a general comment: "The R language can also be embedded into other software packages". There are quite a lot of projects that embed R, mentioning TIBCO and Weka seems a bit out of place. SiggyF (talk) 22:26, 16 June 2010 (UTC)
 * Hi Siggy,
 * I am still on the fence on the subject. What other projects can you mention ?  —Preceding unsigned comment added by Talgalili (talk • contribs) 06:54, 18 June 2010 (UTC)

Split Features
I would like to separate the "Features" section into "Statistical Features" and "Programming Features." The reason is that a computer science student studying computer languages needs a different description R than a student trying to apply statistical techniques as part of a course. The statistics student wants to get to whether R has the method (OLS, ANOVA, GLM, Survival Curves) they need ASAP; while the computer science student wants to classify the language (command line interpreter with functional and object oriented features) and perhaps have a description of data structures (matrices and lists), assignment statement ("<-" instead of "=") conditional statements (IF) and iteration statements (loops). —Preceding unsigned comment added by Jim.Callahan,Orlando (talk • contribs) 17:16, 5 July 2010 (UTC) Jim.Callahan,Orlando (talk)

internal vs external links
I recently remove many external links and replaced many external links with internal links. this was undone in this edit. The edit summary linked to WP:WTAF.

I don't understand this for many reasons. First, the nutshell says, "Editors are encouraged to write the article on a given subject BEFORE adding a link to the article to list pages, disambiguation pages, or templates." but this isn't a list page, a DAB or a template. Second, it makes no references of replacing red internal links with external links.

Most importantly, external links on Wikipedia should generally be in a section at the end, and many of the concepts linked to really should be external links.

Finally, as a general rule, if you don't like part of many edits, you should go through and fix that part. 018 (talk) 18:08, 2 October 2010 (UTC)
 * The issue at hand was the large number of redlinks created by the edits that were reverted. Internal links have no use if the articles they purportedly link to do not exist. Hence the reversion to a version which left functioning links, albeit external. --Alan the Roving Ambassador (talk) 21:23, 2 October 2010 (UTC)
 * The Wikipedia policy that is relevant is WP:EL. I can't find the spot where it says that external links are better than no or red links. Maybe you can point me to it. What I do see is that a link to the official website is allowed and that links that are to material that would be included if the article were of featured article status, would be allowed. Essentially, by saying that you think the external links should eventually go away, you are saying that they should never be there. 018 (talk) 21:54, 2 October 2010 (UTC)
 * Please don't put words in my mouth. What I am saying is that if external links are to be replaced with wikilinks, the articles being wikilinked to need to exist first. Hence my application of WP:WTAF. --Alan the Roving Ambassador (talk) 22:00, 2 October 2010 (UTC)
 * You are saying that the version you prefer is better than the version I prefer. There are three options. (1) largely remove external links from the body, replacing some with red links, (2) largely remove external links from the body, unlinking those that would become red, (3) keep external links in the body. Since the external links are not inline with the policy WP:EL, 1 or 2 are the viable options. WTAF is an essay, what is more, even it claim to no apply in its nutshell text (see my first comment). 018 (talk) 23:06, 2 October 2010 (UTC)
 * Again, you are putting words in my mouth. My statement, verbatim, is that if external links are to be replaced with wikilinks, the aticles being wikilinked to need to exist first. As further reversion to avoid loss of information would potentially place me in violation of WP:3RR, I'm going to step back and ask for an outside opinion. --Alan the Roving Ambassador (talk) 00:06, 3 October 2010 (UTC)


 * If something clearly would be a quality reference, I'm all for adding it, and I think a few of these might fall into that catagory.
 * Just to lay out my agenda, someone nominated this page for GA status. I'm trying to help that along by getting rid of the really obvious policy violations that I couldn't argue IAR against with a straight face. Right now, a huge problem with the article is that there is one secondary source (the NYT article), one blog post and everything else is basically press releases. I don't see how adding a bunch of link to corporate/project pages is going to help that. 018 (talk) 03:56, 3 October 2010 (UTC)


 * I stand corrected on my interpretation of the policies and guidelines in question.
 * I'd be all for seeing some of the redlinked references converted to Template:cite web instances. It would maintain the original (and, subjectively, useful) content, and would also avoid potential issues with new-article creation where another editor may question the relevance of said article and nominate it for deletion under CSD:A7 (although such nominations are often rejected under WP:NOTCSD). --Alan the Roving Ambassador (talk) 23:23, 3 October 2010 (UTC)

Comparo
Might be nice to see a comparison between R and S-Plus in this page. Message me if you would be willing to help put something rudimentary together. Swism (talk) 15:17, 25 February 2008 (UTC) There is an R FAQ on this topic: http://cran.r-project.org/doc/FAQ/R-FAQ.html#R-and-S —Preceding unsigned comment added by 71.235.36.31 (talk) 02:57, 22 November 2010 (UTC)

Number of R packages available
How many R packages are available for free from CRAN? Anwar (talk) 12:30, 12 April 2010 (UTC)
 * Currently, the CRAN package repository features 2322 available packages.
 * http://cran.r-project.org/web/packages/
 * Talgalili (talk) 14:41, 12 April 2010 (UTC)

As of Nov 2010 its between 2300 and 2450 depending on your OS. —Preceding unsigned comment added by 71.235.36.31 (talk) 03:00, 22 November 2010 (UTC)

Statistical Features
I think the description of R needs to be at least as sophisticated as those of other programming languages.

I absolutely agree and support your suggestion for addition to the article. Talgalili (talk) 07:54, 23 November 2010 (UTC)


 * R is designed to be a true computer language? What on earth is a true computer language? If we describe what we mean by true computer language with R in mind, then I think we'll be more successful in elaborating the featues of R.
 * it allows users to add additional functionality by defining new functions - This phrase doesn't mean much and could actually be misleading. We should probably say that users can add functions to R (which is hardly a highlight). In its current form, this seems to somehow mean that being easily extensible is a feature of R, which I would think is hardly the case (at least not case enough to be highlighted in the introduction to R). We should probably just say that R is easily extensible through functions and extensions, and the R community is noted for its active contributions in terms of package. But this point is anyway iterated in the second paragraph.
 * Many of R's standard functions are written in the language (in what language?) should be Many of R's standard functions are written in R itself.

Programming Features

 * When I look at this section, I want to be able to know at once the aspects of R as a programming language. What are the paradigms that R supports? Procudural? Object-Oriented? Functional? And how well does it support those paradigms? How does it do the memory management? Is it interpreted or compiled? A reader shouldn't have to scout through the loose collection of inexpressive text to find out such information, particularly if the heading says Programming Features.
 * As a programming language, R is a command line interpreter similar to BASIC or Python. R is a command line interpreter?! Should be the standard implementation of R (provided by R Development Core Team) comes with a command line interpreter (but other guis exist etc).
 * The above example is deceptively simple because, like APL, R implements matrices, so R can from the command line add or even invert matrices without explicit loops. errr... where is the matrix? and do we really have to say all that to explain how R is great at adding 2+2?
 * A generic function acts differently depending on the type of arguments it is passed. In other words the generic function recognizes the type of object and selects (dispatches) the function (method) specific to that type of object. For example, R has a generic print function that can print almost every type of object in R with a simple "print(objectname)" syntax. why are we describing generic function in so much detail even after the necessary and given wikipedia link? We should simply say that because R has generic functions, print(objectname) can do blah blah blah.
 * Extending R is also eased by its permissive lexical scoping rules. Now this is what should have been described in more detail rather than generic functions. I am not competent to do that though. -- unsigned comment by User:Obelanna 2010-11-09T07:06:04

Obelanna

GUIs
Most people who read this page are not going to care in the least that R can be run inside gretl, for one example. The ever-growing list of GUIs should be moved to the R wiki.--Statr (talk) 20:55, 30 November 2010 (UTC)


 * Hello Statr,
 * I respectfully disagree with your opinion. Since the subject matter is a language for doing statistics, and since the dominant softwares available are GUI based (e.g: sas, SPSS, JMP, excel and so on), I believe there is much of a point to include the GUI perspective of R (especially for the day they will be able to offer a good competition for the commercial products).  As to how this section should be organized, that's a good question for discussion - but should it exist - I think the answer is yes.
 * p.s: I think even fewer people interested in R GUI will visit the R wiki....
 * Cheers, Talgalili (talk) 21:55, 2 December 2010 (UTC)


 * Just try Googling for some of the GUIs. For example, 'nexusBPM' has only 1300 hits, which is not significantly different from 0.  In other words, nobody cares (other than the developer).  Maybe a few of the GUIs are relevant, but most of the ones listed are not.  Remember the target audience--Wikipedia is a general-purpose encyclopedia, not a reference book for R hackers.--Statr (talk) 15:02, 6 December 2010 (UTC)


 * I Statr, I agree with you on that we shouldn't care so much about keeping a link to every R GUI out there. But the major ones should be mentioned (Rcmdr, Deducer), they ARE big.  The "smaller" ones should also be kept (like red-r), since they are "of interest".  Connections to known softwares (like RExcel), I believe are also important.  But you are right that there should be a limit.  The GUI you mentioned is border-line for me (I never heard of it, and don't know how relevant it is).  Do you have a criteria to suggest for who to remove?  (number of google hits is o.k., but I'm not sure where to draw the line).  Best, Talgalili (talk) 17:29, 6 December 2010 (UTC)

R and Scheme
I don't understand the reference to Scheme. R's semantics are derived from S's not Scheme's semantics. Is the sentence in question saying that S's semantics are derived from Scheme's? S and R do indeed support functional programming, but I think it's misleading to say that the semantics of R or S are derived from Scheme. S (and hence R, which is an open source implementation of S) differ from Scheme in so many ways that I think the reference to Scheme should be removed from the page. Do others agree?


 * The main way in which R's semantics are similar to Scheme (and different from those of S-PLUS) is the evaluation model for nested function definitions, as explained in lexical scoping section of the R FAQ. - Avenue 08:23, 4 December 2005 (UTC)


 * Lexical scoping into nested functions is not really what I first think of when someone mentions Scheme. Perhaps Scheme was the first language to do it thus, but these days most languages with nested functions work that way.  Mentioning Scheme gives me the expectataion of, say, having continuations in R, or the lisp-ish "everything is a list" idea, or scheme-like macros.  None of that is in R (as far as I can see).  I think ScheThe me is misleading. Subtilior 20:35, 21 January 2007 (UTC)


 * Saying "everything is a vector" in R might be too strong, but it isn't far wrong. Your other expectations are probably much more of a stretch, but still have an element of truth as far as I understand. Our comment about Scheme is attributed to R's developers, and I think it reflects their position, at least as expressed back in Ihaka and Gentleman (1996): "The resulting language [R] is very similar in appearance to S, but the underlying implementation and semantics are derived from Scheme. In fact, we implemented the language by first writing an interpreter for a Scheme subset and then progressively mutating it to resemble S." Would you be happier if this was discussed later in the article (maybe in the Development section), rather than in the lead paragraph? -- Avenue 14:19, 22 January 2007 (UTC)


 * I guess the "is considered by its developers" clause is enough of a get out. I don't really have a problem. Subtilior 06:31, 5 February 2007 (UTC)

"Everything [in S] is an object." (page 24 Chambers (2008)) "In the S language references to ordinary objects are only through names." which could imply that R does not use numeric pointers (page 24 Chambers (2008)) As with Wikipedia, relying only on names can result in ambiguity or in R's case one name masking another. R and S both use the first names they find, but R has an additional place to look. S looks in the function itself and in the global environment. R adds an intermediate "function environment." One of the practical uses of the function environment is it enables the use of Namespaces. Each package can have its own namespace. See Chambers (2008), "Software for Data Analysis: Programming with R" pages 24, 120, 126 & 460-462 Jim.Callahan,Orlando (talk) 22:45, 9 July 2010 (UTC)

R was initially inspired by Scheme rather than S, see http://www.stat.auckland.ac.nz/~ihaka/downloads/Massey.pdf (p. 16 on "The initial language") on how it initially looked like. Lebatsnok (talk) 12:17, 22 March 2011 (UTC)

Criticism
Tha anonymous criticism about the speed of R is rebutted by the benchmark results at http://www.sciviews.org/benchmark/ This is also explained earlier in the article. The criticisms are thus removed. Den fjättrade ankan 15:25, 31 August 2006 (UTC)

But see http://www.stat.auckland.ac.nz/~ihaka/downloads/JSM-2010.pdf - Ross Ihaka's example with data frames slowing down the speed. + The benchmark you cite has comparisons with matlab and the like but not with e.g. python or java, and definitely not with c and fortran. The latter are relevant because the speed-critical parts of r packages typically get implemented in c (sometimes  in fortran). Lebatsnok (talk) 12:22, 22 March 2011 (UTC)

Naming motivation
The article currently states "It was originally created by Ross Ihaka and Robert Gentleman (hence the name R)..." My understanding was that the name "R" was chosen to parallel "S", which it is a dialect of, roughly speaking. While I suppose both could be true (that is, it is not named "T"), a citation would be nice here. Baccyak4H (Yak!) 15:56, 23 February 2007 (UTC)
 * Better late than never — done ! Schutz 16:10, 31 July 2007 (UTC)
 * Excellent. And what a source too ;-) Baccyak4H (Yak!) 18:30, 21 August 2007 (UTC)

I've understood from Ihaka's papers / slides from his webpage that (a) R started as a separate project, initially unrelated to S, and (b) gradually became more similar to S, and (c) the name referred to Robert and Ross rather than S. But see http://www.stat.auckland.ac.nz/~ihaka/?Papers_and_Talks Lebatsnok (talk) 12:27, 22 March 2011 (UTC)

Advertising?
I've removed Statistica from the list of interfaces to R, since, AFAIK, there are no bridges bewteen these two programs...

If I'm wrong, please revert. —Preceding unsigned comment added by 83.89.137.189 (talk) 16:12, 19 August 2008 (UTC)

Yes there is a bridge Lebatsnok (talk) 12:29, 22 March 2011 (UTC)

The reference to Tibco Spotfire appears (to me) to be advertising. Of more general interest would be that the leading statistical and data base vendors: SAS, SPSS and Oracle have all implemented interfaces to R. Jim.Callahan,Orlando (talk)

GUI vs IDE list?
Hi all,

I'm going through the two lists of IDEs and GUIs and they seemed to be rather mixed one into the other (especially I see GUI items that are only IDEs).

I wonder if these two sections should be merged, and/or maybe have a standard format for both.

What do you think about it??

Talgalili (talk) 13:15, 3 March 2011 (UTC)


 * Agree they need merging. Suggest either a paragraph format (not a list), or perhaps linking to a new page, perhaps something like Comparison_of_statistical_packages Statr (talk)  —Preceding undated comment added 15:08, 28 March 2011 (UTC).

Is this a typo?
The first of the external links reads

"Official website, open source implementation of S"

When I click on it, I'm taken to a site about R. Is the "S" a typo? If not, is there an explanation somewhere in here?


 * It's correct, but misleading. Yes R is an implementation of the S language. Tayste (edits) 02:24, 1 June 2011 (UTC)

Programming features
This entire section seems a bit misleading.

It starts: "As a programming language, R is a command line interpreter". I would say it has rather than is a CLI.

It's compared to Python and BASIC so as an example from the Python page: "Python is an interpreted, general-purpose high-level programming language"

You could say: "R is an interpreted, high-level programming language that offers a command line interpreter" instead for a minor edit. R is already listed on the Interpreted language page so I think it's justified at least partially by that for consistency. However, I think this section could probably benefit from a deeper rework. The fact that R offers a CLI is probably not that important and it's discussion doesn't offer much information over the general CLI article.

Sanjuronyc (talk) 11:33, 1 June 2011 (UTC)

de-facto standard
This claim of R being the de-facto standard for statistics computing is backed up by the article "USING THE R STATISTICAL COMPUTING ENVIRONMENT TO TEACH SOCIAL STATISTICS COURSES" which I have found only two citations from in Google, both in papers using R. Now this is a bit of a strong claim to make and I think it deserves a more credible source. Frenchwhale 16:01, 31 July 2007 (UTC)
 * I'll agree that the statement is too bold with no reliable sources. I've been looking around Google Scholar for any indication of "statistics software" and "standard" or "de-facto", and my results are very mixed. Basically all the statistical programs you can think of (SAS, Microsoft Excel, Sigma Plot, SPSS, Minitab, Stata, as well as S-PLUS and R) have been touted as "standard", which leads me to think that these claims are POVs. I don't think any of these programs can really be considered "de facto" at this moment (except for certain niches, such as biostatistics). I think it is safe to say that R has an expanding user-base as suggested by the posting rate on the help list, and the increasing number of contributed packages. + m t  18:55, 31 July 2007 (UTC)
 * In technical applications, the main reason that R is used is simply because it is free. In bioinformatics, its use is widespread among researchers and the private sector simply because it would be cost-prohibitive for large private organizations (i.e. pharmaceutical companies) to fork out the cash for licences for other programs such as SAS or Matlab. I have no references for this, just years of personal experience working with these people. This article in general is rather unbalanced - for instance there is no discussion of the limitations or criticisms of R (articles on the other software have this). A major fault with R is its inability to efficiently handle for-loops, resulting in the need for clumsy workarounds, something that should be mentioned here. - 52 Pickup 07:20, 15 August 2007 (UTC)
 * The charge that R cannot handle for-loops efficiently is just ridiculous. It is an interpreted language, but that is by design. It is going to be just as efficient with its use of for-loops as any other interpreted language. There is a widespread misconception that the various functional substitutes for for-loops (Such as apply, lapply or sapply) are more efficient but those misconceptions are created by people unskilled in the proper use of the language. The area where efficiency can sometimes be achieved over for-loops with is the proper use of the indexing system. —Preceding unsigned comment added by 71.235.36.31 (talk) 02:54, 22 November 2010 (UTC)
 * Actually, the main reason it's used is that its development (bug fixing, new feature incorporation, etc.) is far faster and more transparent than say the products listed above. Of course, that is a consequence of its open-source status, which also implies its price.  That said however, saying "de-facto standard" or the like without sourcing is original research as well as POV.  Saying "mainstream" or something similar would be more defensible, IMO.  But I have to object to a criticisms discussion for this or even any software (with possible high profile and noteworthy exceptions like Tufte's criticisms of PowerPoint or Fateman's criticisms of Mathematica).  Perhaps limitations, or comparisons with others, may be more appropriate.  Any software product is designed to do certain things, but certainly not all things; saying "well it can't do X" seems highly unencyclopedic. Baccyak4H (Yak!) 12:47, 15 August 2007 (UTC)
 * While I agree with the fact that it is widely used (personnal experience), I think that calling a de facto standard is unjustified. Adding Template:Verify credibility to the source which has not been peer-reviewed and thus only reflects the opinion of its two authors. I don't know if another tag would be more appropriate. If so, please edit it. If no other reference can be found, we should seriously consider removing this claim. Calimo (talk) 17:57, 7 December 2008 (UTC)
 * Here one (peer reviewed?) recently published book that uses de facto in the first paragraph of the book abstract. + m t  07:20, 8 December 2008 (UTC)
 * Another supporting quote, from the "Statistical software" entry in Wiley's Encyclopedia of Statistical Sciences, 2006: "R is currently one of our languages of choice for the development of research-level software." A very similar paper by the same authors seems to be available here.
 * The discussion above does not seem to distinguish clearly between two questions: whether R is a de facto standard for statistical software development, or for statistical analysis. I'd agree with the former, but not (as yet) the latter. Our article makes the former claim, but about S, not just R. (R is one dialect of S). -- Avenue (talk) 22:10, 8 December 2008 (UTC)
 * Good point. I don't think it is even mainstream for non-statistician and non-programmer, who prefer graphical tools such as SPSS or GraphPad Prism (personal experience).
 * Maybe we could remove the controversial de facto and go on with something like R is a mainstream and widely used tool for statistical and data analysis software development[3 or 4 references cited] Calimo (talk) 12:52, 9 December 2008 (UTC)
 * I think the distinction is worth retaining in our article, although it should probably be made clearer. De facto doesn't seem controversial to me regarding statistical software development. Matlab, SAS and Stata are niche players in comparison. I think the articles currently listed on the Journal of Statistical Software web site are telling; one uses Matlab, one uses SAS/IML, and the other five use R. Not only that, but the authors of three of the five do not even feel it's necessary to mention R in the title. This is simply assumed. -- Avenue (talk) 16:08, 11 December 2008 (UTC)

Personally, claiming that something is de-facto standard based on a SINGLE publication is totally wrong. If it is a true de-facto standard, a dozen publications should exist. The publication is a false-positive; there are plenty of statistical tools (including Matlab's statistical toolbox, SAS etc. ) Dimacq (talk) 19:46, 15 November 2011 (UTC)

R as a programming language
From my experience, R is primarily a scripting language. It lacks properly developed features to develop a large codebase; most importantly proper error messages (WITH LINE NUMBERS!) and stack traces (this function called that function and this line caused exception). I think that this should be explicitly mentioned. I know that most of people who are developing this page are very excited about R, but let's try to fairly assess this: it is impossible to write any medium-to-large (>~10k lines) modular, re-usable and maintainable R codebase.

Dimacq (talk) 19:51, 15 November 2011 (UTC)


 * Hmmm, perhaps the short paragraph here - Scripting language - is relevant.
 * Large codebases have been developed with R; Bioconductor is one example. I'm not aware of a facility for attaching line numbers to error messages, but R does have a suite of debugging facilities (see e.g. this overview). They generally support a more interactive approach. The traceback function produces the call stack for the last error. --Avenue (talk) 04:50, 17 November 2011 (UTC)
 * Tal Galili (talk) 17:46, 19 November 2011 (UTC)

You got me there. Traceback is the command I didn't know about. Thanks!!

Linking to an R web interface website - and adding a web interface section
I am copying the comment I got from Slowtortoise4 after removing his link, and would be happy to hear others opinions on the subject.

Slowtortoise4 wrote: You edited the R (programming_language) page and removed a link to a web interface that I added because it was a "not-notable-enough website". To help me understand the rationale, could you tell me why you did not apply the same principle to the other items on the same list, specifically, the link to R AnalyticFlow, Red-R, RKWard and RStudio? In case it was due to language that was not neutral, I've added the link again using the same neutral language as described the other items on the same list. It seems that all these options are equally valid and potentially useful to the public - and should have the same principle applied to all items. Thanks. Slowtortoise4 (talk) 23:48, 6 December 2011 (UTC)
 * I (Tal Galili) wrote:
 * Hello dear Slowtortoise4,
 * First - please know that my editing is NOT intended to offend you in any way. I did it as best as I understood the article.  However, I removed your link again - and would like for us to discuss this matter here before we add the link again :)
 * Second - this entire section about R and IDE/GUI is massed up. RStudio, for example, is both a GUI (see, for example, this project) and an IDE.  In both cases it should be mentioned since they are clearly (in my view) very notable (for example, they were one of the sponsors at the last useR2011 in England). With regards to RedR and R AnalyticFlow, there was a debate awhile ago if to keep them or not, and I voted in favor of them.  My reasoning was (and still is) that they each show a very unique attempt at creating an R GUI, and should be mentioned because of it (even that both do not seem to be widely used).
 * Now let us get to your link -
 * What you write about is (to my humble understanding) NOT a GUI nor is it an IDE. It is simply a new (let us call it) interface to R, through the web.  I imagine it is based on something like RApache or something similar which I may not know about.
 * But in any event it is not a new editor for R, nor is it a new GUI environment for it.
 * If you are interested in including the link (which I imagine you do, since I imagine it is your website), then a section should be added about (for example) web interfaces of R (were the most polite thing would probably be that you would write it, paste it here - and other editors of the page could give an opinion). Such a section should (in my opinion) include mentions of RApache, R-Node, and MAYBE some of their use cases (like your website).  However, it could end up that this section would stay without the use cases, since there are probably many, and mentioning one over the other may not be "balanced". In which case, you can write a new wiki page called "R webinterfaces use cases" - and if it gets to be notable enough to stay on wiki - we can link to it from here.
 * I'd be happy to know what you think, and also what other editors here think. Tal Galili (talk) 06:29, 7 December 2011 (UTC)


 * Thanks for creating this discussion. I'm confused by the reasons that led you to conclude that the proposed website is "NOT a GUI nor is it an IDE ... not a new editor for R, nor is it a new GUI environment for it". The proposed website, is a tool to write R scripts and view both text and graphical output. It allows full use of R including packages. It works within a web browser without software download or installation. Why would it not be an "editor" if a person may write and edit R scripts? Why is it not a GUI if a person may run scripts and view output without manually running the R software at the command line? It does not use R Apache, although the developer of that project appears to have stopped work on that project (I followed Tal Galili's link above). It does not use R-Node. But, it is different in that R-Node must be downloaded and installed, whereas the proposed website tool can be used as is inside a web browser. It is useful students who are less inclined to learn how to install and configure open source software. It is also useful for people who do not have their own computers or servers to download and install software. Both Red-R and R AnalyticFlow, currently on the page, require software download.
 * In response to Tal Galili question, the proposed website uses Javascript JSON and a wiki front end, which provides access to wiki features such as version control, which is useful for learning how to write R scripts.
 * I do not mind if the discussion results in excluding the proposed website tool. I am writing to ask for a fair application of the principles to include or exclude tools. If it is to the benefit of the R community and the public at large to have a variety of alternative front ends to use the R software package, then, it seems like all such tools should be included. If the tools must have an academic reference or other objective indicators of "notability", then, all tools that do not meet that requirement should be excluded. In that case, the website tool I'm proposing should be excluded. But, other tools on the current page should also be excluded because they also do not have an objective indicator of "notability". Thanks. Slowtortoise4 (talk) 09:23, 7 December 2011 (UTC)
 * Dear Slowtortoise4.
 * To my understanding - your web implementation of R is not a new IDE or GUI. It is just a cloud implementation of R with a (very minimal!) front end (which is not "new").  If anything, I would link to projects like RWeb - that also provide the source code for building websites such as the one you implemented.  And again, a more "GUI" website built on R is a website like this:  http://www.stat.ucla.edu/~jeroen/ggplot2/  Which I do not think fits to an IDE or a GUI section (it would, MAYBE, fit to a more general "R in on the web" sort of section).
 * So my position stands as is, and I await the opinions of other editors.
 * Tal Galili (talk) 11:04, 7 December 2011 (UTC)
 * p.s: another example of a web user interface (WUI) of R http://cloudst.at/ or http://www.cloudstat.org
 * I am not sure if we should add a section like this to the article - because since everyone can open such a site - it can easily be that every university who has such a site will put their link here. I am not sure what is the added value in such a case, and what criterion to use.  I am eager to read the opinions of other editors.  Tal Galili (talk) 11:21, 7 December 2011 (UTC)
 * Hi Talgalili. To help me understand your rationale above, could you apply the same rationale to the existing items on the page? Would it be fair to say that Red-R and R AnalyticFlow is worthy because you deem it more difficult to write traditional application software and that you consider "cloud implementation" of software to be trivial and not worthy of mention? Does source code have to be available in order for a tool to be worthy? Is "minimal" a bad thing? What if minimal was on purpose to attract new users to the R community? Could you tell me how those Red-R and R AnalyticFlow meet the "notability" requirement? It would help me when editing pages in the future.
 * Slightly separate, I'd suggest that cloud implementations are not automatically less worthy or trivial. Having access to software without depending on hardware ownership is valuable for many parts of the world. Not every university will provide their own cloud apps because of the security requirements and the costs of the server infrastructure. Besides, if that does come to pass, then, it would be easy enough to change the page at that time. But, it does not seem to be useful to keep some tools and exclude some tools simply on assumptions about whether a tool is difficult to build or not. Thanks. Slowtortoise4 (talk) 17:50, 7 December 2011 (UTC)
 * Hi Slowtortoise4.
 * You are asking wonderful questions, but I am not sure that having me go through and answering them will take us very far.
 * How about we try and see what we can agree upon:
 * Can you agree with me that what you added was a link to a GUI section - where it does not belong? (for example, on the octave page, where you also are having a similar argument, you didn't put it in GUI)
 * If we agree - that means we need to add a WUI/cloud-application section to the article. If this is what you propose, I think it is a good idea.  Please write it and post it here for other to review.  If it is "good" and balanced, people will approve.  I am not sure your link will survive in the long run - but try and we will see where it leads.
 * :) Tal Galili (talk) 19:17, 7 December 2011 (UTC)
 * I would go further and admit that the website I proposed to add does NOT have an objective indicator of notability according to the Wikipedia guidelines. I'm happy for it to be removed from the GNU Octave page, too. But, would you agree that some of the existing tools also do not have an objective indicator of notability and therefore also be removed? I edited the page to add the proposed website because I saw that it was acceptable to have tools such as Red-R and R AnalyticFlow. My objective is not to make a case to add the proposed website. My objective is to fairly apply the same principle to the mention of all tools. Once I was made aware of the Wikipedia guidelines about notability, I thought it was fair to require an academic paper or other objective indicators. What makes it confusing is that the principle seems to be applied in a somewhat arbitrary manner (or at least not transparent to a new user like me). Your views about how Red-R and R AnalyticFlow meets the Wikipedia notability requirements would take us far because it would act as evidence for future users of how the notability principle should be applied. Thanks. Slowtortoise4 (talk) 20:07, 7 December 2011 (UTC)
 * Hi Slowtortoise4,
 * The issue of Red-R and R AnalyticFlow has been discussed here before as others wanted to remove it. I felt like protecting it because I found these projects as very interesting (as in "unusual") at least to me, and that made me feel good with letting them stay (I do not have any other reason, I don't get paid or even know the people behind them).  I agree with you that at the end of the day, many grey areas are encountered when entering the "what link to keep" land.  I think that you won't find a definite answer to what you seek (if there was, we had a bot to do the job, now wouldn't we?!).  But please, I said before that I would be for having a WUI section in R - I would rather develop this idea, then the general "why should one stay and not another".  :)  Tal Galili (talk) 20:15, 7 December 2011 (UTC)
 * I am not interested in promoting a specific website. I believe in transparency and principles. Your most recent reply confirms that personal interests outweigh the Wikipedia notability guidelines. Editors at reference publishing companies are named and accountable for their judgment, indeed, their jobs depend on having their names known. Wikipedia depends on anonymous editors. Guidelines, such as for notability, seems critical to supplement the credibility of Wikipedia. My original proposal was made because I thought that if Red-R and R AnalyticFlow met the notability requirements, then the proposed website could, too. I would not seek an exception to the notability guidelines - I was only seeking fair treatment and transparent application of the notability guidelines. But, it is sad to learn from this first hand experience that the Wikipedia notability guidelines are not applied when something is favored by an editor. Worse, it significantly reduced confidence in the Wikipedia project.
 * I appreciate the opportunity to express my views, but, I do not enjoy conflict situations. This appears to be one person's preference compared to another person's preference. I respectfully withdraw my proposal to add the link to the page. Thanks. Slowtortoise4 (talk) 23:50, 7 December 2011 (UTC)

Tal asked for comments from other editors, so I'll add some comments. I agree completely with Tal that the suggested link was not a GUI (in fact, it rather looks like a terminal window), nor an IDE. I also would have removed the link had Tal not done so. Slowtortoise has an excellent question about what "notable" means. I have removed numerous other references from the IDE/GUI sections as not sufficiently "notable". I have thought about removing more, but I don't know where to draw the line between notable and innovative (but not notable). One crude test is to use a Google search and see how many links are found. Red-R.org has 30,000 links and http://hughesbennett.co.uk/RProject has only 100 links. It would be extremely hard to justify 100 links as "notable". As a point of reference: a Google search for the zenburn color theme has 10000s of links, but the wikipedia page for it was deleted as "not notable". Finally, I think Tal has been very fair and unbiased in this discussion. Statr —Preceding undated comment added 15:38, 9 December 2011 (UTC).
 * Thanks Statr Tal Galili (talk) 17:04, 10 December 2011 (UTC)

I have added a new "useR! conference" section - please let me know what you think
Link: http://en.wikipedia.org/wiki/R_(programming_language)#useR.21_conferences

The section is built upon a combination of these two sources:
 * http://www.warwick.ac.uk/statsdept/user-2011/#About
 * http://en.wikipedia.org/wiki/WordPress#WordCamp_developer_and_user_conferences

Please let me know your thoughts on this matter. Tal Galili (talk) 20:36, 12 December 2011 (UTC)