On Oct 21, 2005, at 8:53 AM, Peter Kleiweg wrote: > James Wettenhall schreef op de 21e dag van de wijnmaand van het > jaar 2005: > > >> We may have to agree to disagree about some things, but I hope >> this has made my point of view a little clearer. >> > > Actually, your elaborate response makes much sense to me. I > understand now that it is not just about replacing the command > line with a GUI. It is not like LaTeX versus Word (i.e. good > versus bad), but about organising and streamlining tasks, doing > "higher level" things. At least, that is what I think it is. > > This is a topic I have been struggling with for quite some time. > For years, I have been working on software for dialectometrics > and cartography. At the beginning, just for doing research at > our institute. But soon, it developed into something people from > other institutes can use. A large set of command line programs, > manual pages, an R interface, and quite an extensive tutorial > with example material. > > My employer urged me to add some sort of GUI. It would make more > people willing to try using the software. I resisted the idea of > a GUI. For one thing, I work on Linux but the GUI should be used > on Windows. (Java is too bothersome. Smalltalk too clumsy. And I > didn't know about Python yet.) But the main problem was: I had > no idea what a GUI should look like, what it was supposed to do. > It took me quite some time, working with my own software, before > I was able to look at it from a distance. The software is just a > toolkit. I didn't want a Graphical Toolkit. What I wanted was > something like a Graphical Project Manager, something task > oriented, with and interactive help system that guides the user > through the work. > > It is still fresh. I haven't had any responses on people using > the GUI, so I don't know yet if this is what people helps. What > I still think as one of the biggest obstacles for using my > software is not cured by the GUI. You still need to select and > prepare the data. If you want maps, you have to provide map > data, in a format the software understands.
For the applications James have in mind, the data format is basically standardized. From a certain level you might think of every observation residing in a separate file (in one out of a couple of different file formats), so all the user has to do is choose "file format" and basically label every file with eg. "control"/"treatment". This is somewhat simplifying of course, but basically the data input step is much more simple than in "general" cases. > This GUI I built is quite specific. It assumes a quite narrow > purpose (though parts of the software can be used independently > for quite other purposes): you start with a set of dialect data, > you do some calculations on that data to make estimates of > differences between dialects, and you visualise these dialect > differences on a geographic map. > > I still don't see anything like that for R. A general GUI for R? > What are the "higher level task" you use R for? It only makes > sense to me if you want to use R in a specific field, such as in > Bioconductor. You build a GUI to that specific higher level > application of R. Yes, and this is exactly what James have been doing. > Or does anyone want to transform R into something like a > spreadsheet program? There are people making a GUI for LaTeX to > make it look like Word, a WYSIWYG, but to me that seems like a > very silly thing to do. Agreed, although I am humble enough to know that there might be sense in doing so from a certain perspective which I am not aware of. Kasper > For those interested, here is my software: > > http://www.let.rug.nl/~kleiweg/L04/ > > And the GUI is here: > > http://www.let.rug.nl/~kleiweg/L04/pyL04/ > > -- > Peter Kleiweg > http://www.let.rug.nl/~kleiweg/ > > ______________________________________________ > R-devel@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel > ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
