[Rkward-devel] Re: directions for Rkward

[Jim Robison-Cox]

Hello Rkward folks,

  I'll introduce myself again.  I am a statistician who is very familiar
with R.  I would not use a GUI for my own work, but would like to help
with development so that it is of use to my students.  I am not a C++ or
Java programmer, but I think I could learn to work with a toolkit like Gtk
fairly quickly.

In response to Thomas's request for opinions,  I do favor a cross platform
development effort.  If this group does a narrow implementation, someone
is going to decide to port that to another platform (MSWindoz), and it
seems better to plan that into the project, imho.  Working with Obversive
seems like a good idea, but I am not involved in hashing out the
programming details.

   I would favor using Gtk, which purports to be cross-platform.  I
realize that Qt is also (and FOX must be, too?).  Duncan Temple Lang, who
is a principal player in Omegahat < www.omegahat.org>  has put together
the RGnumeric interface which is intended for R programmers who want to
access R functions from within gnumeric.  He also has built bindings to
allow the use of Gtk from R and viceversa.  My naive view is that it would
be good to use these tools to build an RGui. I would also call attention
to his Overview paper
  (see links to pdf paper from http://www.omegahat.org/RGtk/index.html)
which begins with a comment that the event loop of R needs to be extended
and generalized in order to utilize queries from Tk, Gtk, Qt, etc.  The R
core team amazes me with how far they have come in a few years, and I'm
sure that they will address this problem.

 I am trying to work thru Duncan's examples.  He points to spreadsheet
apps in Gtkextras and bindings to GtkHTML from R.  It seems that many
pieces are coming into place, and a major effort will be in how to utilize
them.  For instance, I think it is imperative that the code input into R
be saved -- or saveable.  SPSS does this with a syntax window which is
usually hidden, but is accessable.  It's critically important to those of
us with limited memory, because a year or 2 after we do an analysis, we
may need to review it, and we must have the code available so that the
analysis is repeatable.  I know that I will not be able to reproduce a
sequence of button clicks.  That's just one example of numerous design
issues in setting up a gui.

Well, that's my 2 cents worth.
Regards,
Jim

Jim Robison-Cox               ____________
Department of Math Sciences  |            |       phone: (406)994-5340
2-214 Wilson Hall             \   BZN, MT |       FAX:   (406)994-1789
Montana State University       |  *_______|
Bozeman, MT 59717-2400          \_|      e-mail: jimrc at math.montana.edu