[6294] in s-news-athena
re:documentation and ease-of-use
daemon@ATHENA.MIT.EDU (charles@Playfair.Stanford.EDU)
Fri Jan 27 18:58:35 1995
To: s-news@utstat.toronto.edu
Date: Fri, 27 Jan 95 15:43:56 -0800
From: charles@Playfair.Stanford.EDU
I've resisted jumping into this conversation for a bit, but would like
to toss in my two cents. I've been thinking about graphics in S from
the perspective of a user, and think my concerns parallel the help
system issue.
One issue that's arising is what it takes to make good consumer
software is often not itself a good basic research topic. For
example, what I'd like in S is full object oriented graphics which can
be modified dynamically. For instance, I'd like to be able to click
on axis labels and change the font, size, and orientation. (I've
spent far more time fiddling with oma and cex and las than I care too,
and at points have found the easiest way to get what I wanted to be to
edit the raw postscript file!!) I'd believe that these sorts of
graphics interfaces are not of intellectual interest to the AT&T
people as it's very much a solved problem implemented in many
packages. (I hope I'm wrong and will happily stand corrected if so.)
As another respondent noted, S was formerly largely a tool for
researchers, but in it's incarnation as S-Plus has become consumer
software. So there's then the question of who's going to do the work
to put nice front ends on it for the casual user. As someone who's written
and distributed poorly documented S functions, I certainly understand that
the work needed to make a polished finished project is often the thing
that slides in favor of newer substantive research.
I'm quite grateful for and impressed with the work of the Bell Labs
people, but it's quite possible that what's needed to make S-Plus a
good solid product isn't equivalent to what makes a good research
topic for them.
Charles Roosen charles@playfair.stanford.edu
Dept. of Statistics
Stanford U.
