[R] In R 2.14.1, what does wrong sign in 'by' argument mean?

[Ben Bolker]

Petr Savicky <savicky <at> cs.cas.cz> writes:

> 
> On Tue, Feb 21, 2012 at 10:53:44AM -0800, bioinformatics wrote:
> > Here is my code
> > 
 
 [deleting context -- sorry, but posting via Gmane and I have
a short comment to make, so if I don't delete stuff it will
complain ...]

> 
> Hi.
> 
> Try to set options(error=utils::recover) before the run. When
> the error occurs, you can see the values of the variables
> inside the function, where the error occured.
> 
> Hope this helps.
> 
> Petr Savicky.
> 

  'bioinformatics': it's considered appropriate to at least
mentioned that you posted this on Stack Overflow as well:
<http://tinyurl.com/86npzft> I and another commenter there
have spent a fair amount of effort already trying to help.
It would be better to address the comments there instead of
spreading effort across two different mailing lists.

  As is discussed there, part of the problem is confusion
over whether you're starting from a single string (i.e.
length(inputseq)==1 and nchar(inputseq) is a numeric vector
with length 1 and value >1) or a string of single characters
(i.e. length(inputseq)>1 and all(nchar(inputseq)==1)).  Beyond
that, and more critically, your 

    chunk = windowsize[starts[i]:(starts[i]+9999)]

doesn't make any sense if 'windowsize' is a single value.

In other words, it's a little hard to believe that this particular
code chunk ever did something sensible.

  Ben Bolker