RE: [R-gui] Re: [R] Feedback about SciViews?

About this list Date view Thread view Subject view Author view Attachment view

From: Philippe Grosjean (phgrosjean@sciviews.org)
Date: Mon 05 May 2003 - 22:14:01 EST


Message-id: <MABBLJDICACNFOLGIHJOGENNDGAA.phgrosjean@sciviews.org>

Duncan Murdoch wrote (about a suggestion to make a java R GUI):

>I think the issue is that it's hard now to write any GUI at all,
>because too much UI is in R itself. As we move towards a separation
>of UI and computation, it may well make sense to write a GUI in Java.

Yes, I fully agree. This is indeed the main, and leading idea of the
SciViews project: a separation of the frontend and the backend.

Otherwise, I was interested by the way this thread is deviated from its
orginal question about impression on the preview version of SciViews (RE:
[R-gui] Re: [R] Feedback about SciViews?). Now, we are once again discussing
which language to use to make a R GUI. I say once again, because it is at
least the fourth, or the fifth time we have this discussion. Well, if there
is something new, where is the problem? Indeed, there is not much new
ideas... nor much more people that move from just giving their own
impression to a more active behaviour. I understand that, because starting a
large project like a R GUI is a big task, and not everybody is willing to
spend a large part of its time and energy to coordinate such a project.

OK, that said, you must realize that, in our (almost) perfect community of
voluntair developers that makes the success of R, the initiation of a R GUI
project is not only a matter of deciding which tool to use in a global and
free discussion within the community. It is rather more fueled by one, or a
few people that actually decide to sacrifice a large amount of their time to
start and coordinate a project. Initiation of such a project is depending on
various criteria, among them:

1) the time the project starts (languages and graphical toolkits are
evolving rapidly, so, it is almost certain that an option taken today would
not necessarily be the same choice if it was done tomorrow; however, when a
project actually starts, many decisions have to be taken, and not delayed
till tomorrow!),

2) the skills, the tastes and the working habits of the people that initiate
a project.

3) the needs, I mean, someone that starts a project does it because he feel
there is a lack somewhere and its intention is to fill the gap. For R GUI,
early discussions last year ended with a conclusion that a
platform-independent GUI should be preferred, but if there is too much
difficulties to reach this independence, then, Windows users should be
targetted first (I was surprised at that time that Macinthosh users where
not reacting,... but at least they do from time to time).

Keep this in mind.

Also, I would say that, if R GUI projects are not evolving rapidly today, it
is precisely because many developers are still wandering which major choices
(like language, graphical toolkit, way to interface R) they would take.
Please, realize now that some people one moving forward. There ARE R GUI
projects in course: SciViews, ObveRsive, Poor Man's GUI, StatPaper,... (see
http://www.r-project.org/GUI). This implies these major choices have already
been done, and also for some projects, it means that several hundreds, or
even thousands of hours of work have been already invested. Please, respect
this. From now, there are two choices:

1) You help in a constructive way these project to move forward,

2) or, you are really unhappy to the choices they did, and more importantly,
YOU EXACTLY KNOW WHY (that is, having a vague idea on what should be good is
one thing, but are you really, really sure you master all the problems?
Certainly not, until you actually start designing and coding the
application!). This means also that, before critisizing choices made in the
current projects, or proposing alternatives, you must really have the same
skills as people that already worked the question of making a R GUI... Well,
fortunatelly, there are several people like that in our community.

To conclude this very long mail (sorry)! I ask the question again: I am
really interested by constructive suggestions and remarks about the first
implementation of a R GUI, as it is done in the SciViews project
(Insider)... and keep in mind that the main objectives of SciViews are:

1) Investigate a separation of the backend and frontend, while offering very
robust interprocess communication and control and,

2) Design an original GUI that is as much ergonomic, powerful, efficient,
and pleasant to use as it could be in the field of statistical and data
analysis.

The fact that it is a Windows-dependent solution is, I believe, less
important that the solutions it could provide to these two topics, and that
could be transposed into other systems/languages, when thay will be
optimized.

Best,

Philippe Grosjean

...........]<(({?<...............<?}))><...............................
 ) ) ) ) )
( ( ( ( ( Dr. Philippe Grosjean
 ) ) ) ) )
( ( ( ( ( LOV, UMR 7093
 ) ) ) ) ) Station Zoologique
( ( ( ( ( Observatoire Oceanologique
 ) ) ) ) ) BP 28
( ( ( ( ( 06234 Villefranche sur mer cedex
 ) ) ) ) ) France
( ( ( ( (
 ) ) ) ) ) tel: +33.4.93.76.38.18, fax: +33.4.93.76.38.34
( ( ( ( (
 ) ) ) ) ) e-mail: phgrosjean@sciviews.org
( ( ( ( ( SciViews project coordinator (http://www.sciviews.org)
 ) ) ) ) )
.......................................................................

______________________________________________
R-help@stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help


About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.3 : Tue 01 Jul 2003 - 09:11:45 EST