Hi Philippe, I disagree - with the new RGtk2 package (http://ggobi.org/rgtk2/) it is very easy to work cross platform with an attractive GUI. RGtk2 works on Windows, Mac and Linux and (in my opinion) is much more aesthetically pleasing the Tcl/Tk.
Rgtk2 supports libglade so if you are familiar with glade, it is a very easy transition, and it is very easy to port work done in other programming languages if you are comfortable converting your (eg.) python code to R. Hadley On 2/19/06, Philippe Grosjean <[EMAIL PROTECTED]> wrote: > Hi Luca, > > I would suggest to work/enhance existing code rather that reinventing > the wheel. There are a couple of RGui projects, and you may found what > you are looking for in there. Look, at http://www.sciviews.org/_rgui/ > for a listing of various projects. You have GUIs written in Java, > Tcl/Tk, GTK, etc... there. Tcl/Tk is often used for convenience, because > the tcltk R package is available with (almost) all R distributions. So, > you don't need to install additional stuff. That is what makes Rcmdr so > easy to run on all platforms. If you want more advanced features, I > agree that there are better solutions (GTK, wxWidgets, etc). > > May I suggest you to look here: > http://bioinf.wehi.edu.au/folders/james/wxPython/. This is a solution > using R and Python... in development and desperately seeking for a new > maintainer. This is not GTK, but wxWidgets, which is also a very good > solution. Regarding a substitute for Glade, you could consider Boa > Constructor. Look here: http://bioinf.wehi.edu.au/folders/james/RBoa/. > The explanation on this page is a little outdated. The latest version of > Boa Constructor is now compatible with wxPython 2.5.X. So, you don't > need to install Python/wxPython twice as explained in the page. Anyway, > using RSPython, you should be able to do what you want (mixing R, Python > and GTK). > > Best, > > Philippe Grosjean > > Luca Manini wrote: > > Hi, > > > > I'm new to this list (and to the whole R world); I've started to read > > some threads in the archives to get acquinted with the community but I > > have some questions ready to be asked "now". So please keep with me > > even if the mail is not that short. > > > > > > I'm a software developer and I've been asked to "write a GUI for R". > > The customer(s) are (could be in the near future) some departments of > > the local university where R is used both for teaching and research > > (in various areas). So the problem is always the same: users find CLI > > difficult to learn/use and/or the time to learn them is greater than > > the total available time (for introductory courses in statistics, for > > example). > > > > As I said, I don't know much of R but it clear that is a big > > language/environment and that just thinking (let alone writing) a > > "full GUI" for it is a major undertaking well beyond the time (and > > money) budget I have at hand. > > > > So, what I have in mind is, instead of using/customizing big GUIs, to > > just write some small apps with a dedicated and minimal GUI for any > > single and simple task (think for example of some students' > > "exercises" on regression, or a series of computations or analysis or > > drawing steps needed to complete a research paper). > > > > Of course I will not directly write them myself (sorry to say that) > > but instead I have to check whether that approach is feasible and, if > > it is, to "set up" the environment (docs, tools, examples, ...) so > > that "junior programmers" (or smart users) can write the single apps > > in a reasonable time. > > > > I'm thinking about Python + GTK + Glade for the interface stuff with > > some "glue" to get Python speak to R (and trap the answers). Python > > is my preferred language, it is easy to learn (and I'm happy to teach > > it) and Glade is easy to use. > > > > My biggest concern at the moment is to check that I'm not offering to > > write something to solve an already solved problem and **for that** > > I'm asking this list's help. > > > > TIA for any help/suggestion, Luca > > > > PS: the second biggest is to check that my idea is feasible/reasonable > > (and for that I've already started experimenting). > > > > _______________________________________________ > R-SIG-GUI mailing list > R-SIG-GUI@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/r-sig-gui > _______________________________________________ R-SIG-GUI mailing list R-SIG-GUI@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/r-sig-gui
