Skip to content

Small R documentation bug in ?anyDuplicated

2 messages · Bert Gunter, Martin Maechler

Original
Bert Gunter
# 
All:

I did not want to bother R folks for an R Bugzilla account, so I'll just
note what appears to be a documentation bug here

In R version 4.0.3, ?anyDuplicated  says:
"anyDuplicated(.) is a ?generalized? more efficient shortcut for
any(duplicated(.)).

However, anyDuplicated returns an integer and any(duplicated)  -- that is
any() -- returns a logical.  I was bitten by this by a function expecting a
logical from anyDuplicated.

I realize that there are scare quotes around "generalized", which would
indicate that anyDuplicated and any(duplicated(.)) aren't identical.
However, I believe that the line above (especially 'shortcut') leads one to
expect that both return a logical. A simple addition such as: "Note,
however, that any() returns a logical and anyDuplicated's  returns an
integer' would avoid the confusion.

Best to all,
Bert
Martin Maechler
# 
> All: I did not want to bother R folks for an R Bugzilla
    > account, so I'll just note what appears to be a
    > documentation bug here

    > In R version 4.0.3, ?anyDuplicated says: "anyDuplicated(.)
    > is a ?generalized? more efficient shortcut for
    > any(duplicated(.)).

IIRC, I wrote that text there ..

    > However, anyDuplicated returns an integer and
    > any(duplicated) -- that is any() -- returns a logical.  I
    > was bitten by this by a function expecting a logical from
    > anyDuplicated.

well, I was very aware / conscious of that and here's why I even
more liked the term "generalized" :

I'd argue that in S and hence R,
{0,1}  are "almost the same" as {FALSE,TRUE},
(they are internally coded as integer 0, 1)
and e.g.
[1] TRUE
> I realize that there are scare quotes around
    > "generalized", which would indicate that anyDuplicated and
    > any(duplicated(.)) aren't identical.  However, I believe
    > that the line above (especially 'shortcut') leads one to
    > expect that both return a logical. A simple addition such
    > as: "Note, however, that any() returns a logical and
    > anyDuplicated's returns an integer' would avoid the
    > confusion.

anyDuplicated() is mostly useful for   if(.)  or  while(.)
and for these, it *is* a shortcut.

    > Best to all, Bert


I really think help pages should be allowed to use expressive
language at times.. notably in the section 'Description'.

There's the 'Value:' section which one should *really* read as
well where it clearly says that anyDuplicated returns the index
in case there are duplicates.

To such a bug report  I'm pretty strongly replying
   "Works as documented"

Martin