[tech-spec] Re: Nested Loops

  • From: Steve Wisdom <swisdom@xxxxxxxxx>
  • To: tech-spec@xxxxxxxxxxxxx
  • Date: Tue, 16 Nov 2004 15:39:40 -0500

BBands <bbands@xxxxxxxxxxxxxxxxxx> : 

> I find the concept of implicit loops being "better" 
> than explicit loops disturbing.

I find the concept of multiple inheritance disturbing, but to each his own 

To my taste the APL'ness of R is its strongest point.  That's the
language's "karma".  Worth noting that FORTRAN (yes, it still exists)
has incorporated vector/matrix'wise aspects in recent specifications,
perhaps in imitation of S+/R, Matlab et al

But it does swap "off by one" errors for "mis-shapen" and "forgot
drop=F" errors, in my experience, and it's unclear whether net errors
are reduced

And I agree that in undocumented R code it's hard to follow the
"shapes" of the objects as they're subscripted, reshaped, outer'ed &

Also, let me shout out to Hawaii, great work on getting your toes in
the water.  And I agree with Oregon's remarks that R is better suited
to "gross" data-smashing than Tradestation-workalike projects
(although I do find myself doing "stupid pet tricks" with R just
because it's handy)

If I were less ignorant I'd write an essay on how to mix & match
ts/its/zoo/Rmetrics to best effect.  "One of these days".  Generally
I'd advise avoiding packages altogether whilst learning the language;
there's plenty to digest in vanilla R.  But price-data wrangling lends
itself to these packages

Other related posts: