[R] extra .

[William Dunlap]

> -----Original Message-----
> From: Steve Lianoglou [mailto:mailinglist.honeypot at gmail.com] 
> Sent: Friday, August 21, 2009 9:02 AM
> To: William Dunlap
> Cc: kfcnhl; r-help at r-project.org
> Subject: Re: [R] extra .
> 
> Hi,
> 
> This is somehow unrelated, but your answer brings up a question that  
> I've been curious about:
> 
> On Aug 21, 2009, at 11:48 AM, William Dunlap wrote:
> 
> >
> >> -----Original Message-----
> >> From: r-help-bounces at r-project.org
> >> [mailto:r-help-bounces at r-project.org] On Behalf Of kfcnhl
> >> Sent: Thursday, August 20, 2009 7:34 PM
> >> To: r-help at r-project.org
> >> Subject: [R] extra .
> >>
> >>
> >> sigma0 <- sqrt((6. * var(maxima))/pi)
> >>
> >> What does the '.' do here?
> >
> > In R it does nothing: both '6' and '6.' parse as "numerics"
> > (in C, double precision numbers).  In SV4 and S+ '6' parses
> > as an integer (in C, a long) and '6.' parses as a numeric,
> > so putting the decimal point after numerics makes the
> > code a bit more portable, although there are not too many
> > cases where the difference is significant.   Calls to .C, etc.,
> > and oveflowing arithmetic are the main problem points.
> >
> > In R and S+ '6L' represents an integer.
> 
> If this is true, I'm curious as to why when I'm poking through some  
> source code of different packages, I see that some people are 
> careful  
> to explicitly include the L after integers?
> 
> I can't find a good example at the moment, but you can see one such  
> case in the source to the lm.fit function. The details aren't 
> all that  
> important, but here's one of the lines:
> 
>      r2 <- if (z$rank < p)
>          (z$rank + 1L):p
> 
> I mean, why not just (z$rank + 1):p ?
> 
> Just wondering if anybody has any insight into that. I've 
> always been  
> curious and I seem to see it done in many different functions and  
> packages, so I feel like I'm missing something ...

Some developers are more fastidious than others.  Since lm.fit()$rank
is an integer and the sum must be integral it make senses to add an
integer instead of a numeric so you are passing an integer to colon.
The colon operator will return an integer vector in either case but it
seems cleaner (and very slightly faster) to pass it an integer.

Bill Dunlap
TIBCO Software Inc - Spotfire Division
wdunlap tibco.com 
 
> 
> Thanks,
> 
> -steve
> 
> --
> Steve Lianoglou
> Graduate Student: Computational Systems Biology
>    |  Memorial Sloan-Kettering Cancer Center
>    |  Weill Medical College of Cornell University
> Contact Info: http://cbio.mskcc.org/~lianos/contact
> 
>