[R] off-topic: better OS for statistical computing

Patrick Connolly p_connolly at ihug.co.nz
Tue Sep 11 12:48:34 CEST 2007


On Mon, 10-Sep-2007 at 07:26PM -0400, Gabor Grothendieck wrote:

|> My sense is that R users are even split between UNIX and Windows
|> users so either will do in terms of the larger community.
|> 
|> Some R packages may not be avaliable on every platform or will
|> be available on one platform before another or there will be
|> certain platform-specific issues.  So in the end its easiest to
|> have the same thing everyone else that you work with does.

That sounds like a reason to have *more than* one.  If you burn
bridges to one OS, there'll be cabability you're shutting yourself off
from.  If the idea of uniformity is one of equally disadvantage, I
suppose you're right.

It's attractive for administrators to have as little diversity as
possibile because it makes their job easier.  Imagine what it would be
like if statisticians wielded power and required all data to be in a
form that they could analyse using a simple anova.  Clients would have
their nice little anova tables with LSDs so they could tell what was
significantly different from others.  No need to learn how to read a
different type of output.  Nice and simple.

It's not an approach I would go along with.  It would make the
statistician's jobs easier -- though it would also be very boring.
Then there would be the effect on those who had reason to use very
different data structures....

[...]

|> Also associated software may be, for example, Microsoft Office in a
|> Microsoft environment and LaTeX in a UNIX environment. And
|> networking will be simplified in a consistent environment too.
|> Certainly there is Open Office, Samba and putty but the easiest is
|> just not to have to worry about getting everything to work together
|> by just having the same thing in the first place.

This is the Anova Everything approach again.  


|> 
|> Neither Linux nor Windows is superior to the other.  People
|> making such representations generally know one much better
|> than the other and its more a reflection of their own experience
|> than anything else.  

That's an interesting observation.  My experience is very different.
Nearly everyone I know who knows how to use Linux also knows more
about Windows than most people who've only ever used Windows.  (I wish
I had a dollar for every Windows user I've shown how to change the
vertical refresh rate on their monitor.)

[....]

There are organisations for whom Macintoshes are best; for some it's
Windows; for others it's Linux or Unix.  Those with a variety of
activity do themselves a favour by having multiple platforms.  The
extra effort put into their computer services people pays off in
productivity (if they're interested).

I would venture to say that many users of R would be involved in that
type of organisation.  The original poster might be in that position.
My comment was to help counter claims given for the purported need for
hegemony.

One of the many great things about R is it's cross-platform
capabilities.  That wasn't achieved using a monolithic approach to
software.



-- 
~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.   
   ___    Patrick Connolly   
 {~._.~}          		 Great minds discuss ideas    
 _( Y )_  	  	        Middle minds discuss events 
(:_~*~_:) 	       		 Small minds discuss people  
 (_)-(_)  	                           ..... Anon
	  
~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.



More information about the R-help mailing list