R devel - Jan 2005 - mydataframe$colname: using substring of colname may also match some column (PR#7496)

Full_Name: Marc Mamin
Version: 1.8, 2.0.0
OS: Windows & Linux
Submission from: (NULL) (217.17.202.254)


Using only the beginning of a column name will match it:
>aaa<-1
>df<-as.data.frame(aaa)
>names(df)
[1] "aaa"
>df$a   
[1] 1 !!!!!!!!! (I expect df$a to be undefind)
>df$x
>NULL
Compare with:
> df["aaa"]
  aaa
1   1
> df["a"]
Error in "[.data.frame"(df, "a") : undefined columns
selected


Here another example that underline how problematic this issue can be:

aa1<-1
aa2<-2
df<-as.data.frame(aa1,aa2)
> df$aa
[1] 1   (only the first matching column is retrieved)

That is the documented behaviour, not a bug!

BTW, df is the F density function.

If you want an exact match, use  df[match("aa", names(df))]

On Wed, 12 Jan 2005 marc@intershop.de wrote:
> Full_Name: Marc Mamin
> Version: 1.8, 2.0.0
> OS: Windows & Linux
> Submission from: (NULL) (217.17.202.254)
>
>
> Using only the beginning of a column name will match it:
>
>> aaa<-1
>> df<-as.data.frame(aaa)
>> names(df)
> [1] "aaa"
>
>> df$a
> [1] 1 !!!!!!!!! (I expect df$a to be undefind)
Where did you read that?
>> df$x
>> NULL
>
> Compare with:
>
>> df["aaa"]
>  aaa
> 1   1
>
>> df["a"]
> Error in "[.data.frame"(df, "a") : undefined columns
selected
>
>
> Here another example that underline how problematic this issue can be:
>
> aa1<-1
> aa2<-2
> df<-as.data.frame(aa1,aa2)
>> df$aa
> [1] 1   (only the first matching column is retrieved)
-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

At Wednesday 03:50 AM 1/12/2005, marc@intershop.de
wrote:
>Full_Name: Marc Mamin
>Version: 1.8, 2.0.0
>OS: Windows & Linux
>Submission from: (NULL) (217.17.202.254)
>
>[snipped issues previously responded to]
>
>Here another example that underline how problematic this issue can be:
>
>aa1<-1
>aa2<-2
>df<-as.data.frame(aa1,aa2)
> > df$aa
>[1] 1   (only the first matching column is retrieved)
Actually, the above commands constructed a data frame with one column (the 
second argument to 'as.data.frame' is 'row.names'):

 > aa1<-1
 > aa2<-2
 > df <- as.data.frame(aa1,aa2)
 > df
   aa1
2   1
 > dim(df)
[1] 1 1

When a data frame (or list) does have columns whose names have common 
prefixes, the behavior is as documented (NULL is returned):

 > df2 <- data.frame(aa1, aa2)
 > df2
   aa1 aa2
1   1   2
 > df2$aa
NULL
 >