[R-gui] Editor in R and Windows focus problems

Byron Ellis bellis at hsph.harvard.edu
Thu Apr 29 16:51:47 CEST 2004

I have to disagree with the assessment that identical interfaces across 
platforms is the optimal situation. IMHO, the Windows version should 
act like a Windows program, the Mac version like a Mac program and the 
UNIX version like a UNIX program and so on. wxWindows makes the best of 
the bad situation, but its still a local optimum at best. Better to get 
R's core functionality out of the event loop business entirely and 
switch to an embedding model---including moving graphics devices out of 
the core and into packages (which seems to be happening slowly of its 
own accord). Now that I'm thinking, a C-level interface to library() 
might be a handy idea (s.t. a GUI could load GUI specific packages, 
such as graphics devices) towards that end.

I realize it sounds like I'm advocating interface anarchy. I am to some 
extent, but this could be mitigated by the establishing of an R Human 
Interface Guidelines (HIG) a la, Microsoft or Apple or Red Hat. This 
could mandate certain things such as menu items while still allowing 
platform specific discretion (a good example might be preferences---a 
Windows preferences interface is much different that its OS X 
counterpart) and features. Conveniently we seem to have a mailing list 
for this sort of thing right here. ;-)

On Apr 29, 2004, at 10:20 AM, Dirk Eddelbuettel wrote:

> John,
> On Thu, Apr 29, 2004 at 10:15:10AM -0400, John Fox wrote:
>> Dear Philippe, Duncan, et al.,
>> I hesitate to volunteer other people's work, but I wonder whether it 
>> isn't
>> time to think about replacing the GraphApp-based Rgui as the standard
>> interface to R under Windows. In particular, it occurs to me that a
>> scaled-down version of Philippe's SciViews -- including a basic set of
>> menus, the Syn text editor, and perhaps an object browser -- would 
>> provide a
>> superior interface. Of course, there may be obstacles that I don't
>> understand.
> That won't work as it is Windows-only. IIRC even Philippe now 
> acknowledges
> that this is suboptimal.
> AFAICT the emerging consensus has been for quite some time to move to
> wxWindows (now called wxWidgets). But someone needs to do the work...
> Dirk
> -- 
> The relationship between the computed price and reality is as yet 
> unknown.
>                                              -- From the pac(8) manual 
> page
> _______________________________________________
> R-SIG-GUI mailing list
> R-SIG-GUI at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/r-sig-gui
Byron Ellis (bellis at hsph.harvard.edu)
"Oook" -- The Librarian

More information about the R-SIG-GUI mailing list