Date: Wed, 26 Feb 1997 16:08:04 -0500 From: Russell Almond Here are a couple of the major differences: 1) Graphics Engine. One of the biggest drawbacks of S is that it still has a plotter based graphics package. I find S easier to use for fancy static graphics with lots of overlays. Xlisp-stat's big advantage comes from its ability to support dynamic graphics. 2) Memory model S stores data objects on disc while XLS stores them in memory. One consequence is that S storage is persistent. XLS should also be faster with some operations as a consequence, but S will handle larger data sets (although with the price of memory, thats not as much of a problem as it used to be). 3) Programming Ease I'd say this one is a toss-up. S has a more C-like syntax for those who have a mental block about parentheses. S also has a truely bizarre object model (class identity and method dispatch are completely divorced from object encapsulation.) Also, I find the S lazy evaluation harder to understand than the Lisp eager evaluation rules. Unfortunately, xlisp lacks many of the debugging tools which makes Lisp so easy to use. (Which reminds me, did I ever post my debugging code to the list?) 4) Functionality S-Plus definetly has more, and more thorough testing. Both collections of functions reflect the research interestsof their respective authors: S tends to favor robust procedures and XLS Bayesian. On the other hand, if you really need a specific function, try SAS. 5) Support Luke does a great job for a one man show, but part of the price of S-Plus is the ability to call somebody on the phone to hold your hand while you go through the mysteries of configuring S-Plus to work on your PC with your printer. Also, there are training couses available. 6) Price XLS costs fewer dollars, but morein sweat equity. Thus, it tends to be favored by researchers and teachers who are willing to put in the effort to get it to do what they want. --Russell Almond --------------------------------------------------------------------- From: Jan Deleeuw Subject: Re: S-plus and xlisp-stat Date: Wed, 26 Feb 1997 13:31:28 -0800 (PST) Generally, I think the comparisons of S-plus and XLS have been fair. Also, there is one in Statistical Science some time ago. Although we have both, the five major and overriding reasons why I would not even consider using S-plus are perhaps worth repeating. (a) XLS comes with complete source code -- in ANSI C. Generally it is a breeze to port it to a new flavor of UNIX. It is quite easy to extend it at compile time, and by now it is also easy to extend it at runtime with object code. One can always go back to the source in case of doubt. One can look at the source and marvel. (b) XLS runs on the Mac, in fact that is where both XLS and I prefer to be most of the time. The other two are just necessary evils. (c) I can give it away to my students and my family, in unlimited quantities, and for their own home computers, no matter what hardware and OS they have. I can provide WWW interfaces to the program which run for free from the web servers -- I dont think the S-plus people would appreciate a similar service. (d) Porting from Common Lisp is now easy. Porting from FORTRAN to CL, and then to XL is now easy. Lisp is a general purpose language. There are lots of books on Lisp. Lisp is a beautiful language. Because of XLS I now read books such as "On Lisp" or Norvig or Steele. (e) Dynamic graphics for demos, instructional modules, and research tools cannot really be done in S. Computing and making static plots is perhaps marginally easier in S, but this is not what I do. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw