To all: In partial defense of S-Plus and to (I hope) stimulate thoughtful
responses.
A few points:
1. The view expressed below by Harvey Monder is harsh, but makes at least
one valid point: S is a complex language that makes a lot of demands of the
user. I would suggest that at least some of the problems experienced by
Sicco Schets may, in fact, not be problems at all, but simply a result of
his inexperience with the language(as he, himself, acknowledged). That is,
the usage IS documented and correct but he has simply not understood it
fully. Of course, there are problems, and I do not mean to dismiss
everything he wrote. I leave it to those more expert than I to sort things
out.
2. That said, what Prof. Ripley finds clear and straightforward should not
be used as a guide for what the rest of us should find clear and
straightforward. Given S-Plus's complexity, clear, effective documentation
is vital -- especially for applications oriented users who have no need to
engage in S's serious software development capabilities. I think that there
is broad agreement even among those of us who find S indispensable that
documentation is one of S-Plus's weaker points. I know that there are good
reasons for this: given limited resources and the demands of the
marketplace, documentation usually takes a back seat to feature development.
I know that Insightful is trying to maintain a better balance here, but they
clearly have room for improvement.
3. One way Insightful has tried to provide functionality to the
applications-oriented user (who are probably the bulk of S-Plus's intended
market) is through the S-Plus GUI. However, many of us have found the GUI to
be more of a problem than a help. This is certainly a matter of personal
style, and I do not mean to imply that many others have not found it to be
key to accessing S-Plus's extraordinary capabilties. But many of us prefer
the language's inherently greater flexibility and power even if we don't
develop software. This means that the language has to serve both the needs
of "casual" appliations-oriented users as well as those who want to write
serious data analysis software applications, for which, I think, S is
without peer (nothing else even comes close!). The recent controversy about
the new class system highlights the difficulty here: it may be that the
changes that improve the languages capabilities for serious developers may
make it more difficult for casual users. Again, this is a difficult balance,
and I hope that Insightful will continue to look at this issue with the
intention of possibly providing tools or modifications IF (and I emphasize
"IF") this does, indeed, turn out to be a problem for a significant number
of folks.
4. In the end, of course, one's choice of what to use for data analysis
software is a personal decision that depends on one's individual needs and
capabilities. However, I think it fair to say that any serious data analyst
who dismisses S-Plus because of a few comments on this or any other list or
even because of new release bugginess does themselves a disservice. Many of
us "ordinary" S-Plus users have found that the language's extraordinary
power, the tremendous graphics capabilities, and the enormous range of data
analysis methodology that is either built-in or available from
user-contributed libraries -- much of which cannot be found anywhere else --
makes the effort worthwhile, despite the occasional frustrations. May I also
say that the user community on this list is remarkably positive and helpful;
having statisticians like Professors Ripley, Venables, Harrell (or
non-Professor Therneau) regularly monitor and offer help to users is simply
fantastic.
So please consider S-Plus carefully. Finally, you should take advantage of
resources beyond that immediately provided by Insightful and this list --
especially the two books by Profs. Venables and Ripley - Modern Applied
Statistics with S-Plus(MASS) and S Programming. I have found the latter to
be especially helpful in my own struggles with the language.
Best regards to all,
Bert Gunter
Biometrics Research RY 70-38
Merck & Company
P.O. Box 2000
Rahway, NJ 07065-0900
Phone: (732) 594-7765
mailto: bert_gunter@merck.com
"The business of the statistician is to catalyze the scientific learning
process." -- George E.P. Box
-----Original Message-----
From: Monder, Harvey [mailto:Harvey.Monder@pharma.com]
Sent: Thursday, September 20, 2001 9:23 AM
To: 'Sicco Schets'; Prof Brian Ripley
Cc: 's-news@wubios.wustl.edu'
Subject: Re: [S] Differences between UNIX and PC version of Splus
As far as I am concerned, unless the documentation issues in Splus are
improved and the backward compatibility and cross platform compatibility are
addressed (at least via documentation) I can not consider Splus as a serious
programming language.
|