[Rd] how to get old type.convert() numeric behavior?

[Andrew Piskorski]

Regarding this change:

> CHANGES IN R 3.1.0: 
>   NEW FEATURES: 
>     * type.convert() (and hence by default read.table()) returns a 
>       character vector or factor when representing a numeric input as a 
>       double would lose accuracy.  Similarly for complex inputs. 
>  
>       If a file contains numeric data with unrepresentable numbers of 
>       decimal places that are intended to be read as numeric, specify 
>       colClasses in read.table() to be "numeric". 

How do I get the old behavior where type.convert() automatically
converts to numeric if suitable, regardless of whether or not the
string has more than 17 digits of accuracy?

Sure, I could first pass every single column of data through a kludgy
checking function like my.can.be.numeric() below, and then set
colClasses to "numeric" or not based on that, but is there a better
way?


my.can.be.numeric <- function(xx) { 
   old.warn <- options(warn = -1) 
   on.exit(options(old.warn)) 
   (!is.na(as.numeric(xx))) 
} 

Example of the changed behavior in R 3.1.0 vs. earlier versions, both
with options("digits"=10) set:
  
# R version 3.1.0 Patched (2014-04-15 r65398) -- "Spring Dance" 
# Platform: x86_64-unknown-linux-gnu/x86_64 (64-bit) 
> type.convert(paste("0.", paste(rep(0:9,3)[seq_len(17)],collapse=""), sep=""), as.is=TRUE) 
[1] 0.01234568 
> type.convert(paste("0.", paste(rep(0:9,3)[seq_len(18)],collapse=""), sep=""), as.is=TRUE) 
[1] "0.012345678901234567" 

# R version 3.0.2 Patched (2013-10-23 r64103) -- "Frisbee Sailing" 
# Platform: x86_64-unknown-linux-gnu/x86_64 (64-bit) 
> type.convert(paste("0.", paste(rep(0:9,3)[seq_len(17)],collapse=""), sep=""), as.is=TRUE) 
[1] 0.01234568 
> type.convert(paste("0.", paste(rep(0:9,3)[seq_len(18)],collapse=""), sep=""), as.is=TRUE) 
[1] 0.01234568 

-- 
Andrew Piskorski <atp at piskorski.com>