R lists -- how are element names handled?

Question

I have noticed this unexpected feature:

foo <- list(whatever=1:10)

Now, the following works as well:

foo$wha
foo$w
foo$whateve

However, the following does not:

foo[["wha"]]

This has the unexpected consequences (unexpected for me, that is) that if you have two potential names, like "CXCL1" and "CXCL11", and you want to know whether CXCL1 is not null by inspecting !is.null(foo$CXCL1), it will return TRUE even if CXCL1 null, but CXCL11 isn't.

My questions are:

1. How does that work?
2. What is the difference between foo$whatever and foo[["whatever"]]?
3. Why would anyone want this behavior and how do I disable it?

Answer 1

1. Partial matching only works with unique initial subsequences of list names. So, for example:

   > l <- list(score=1, scotch=2)
   > l$core #only INITIAL subsequences
   NULL
   > l$sco #only UNIQUE subsequences
   NULL
   > l$scor
   [1] 1
   

2. Both [[ and $ select a single element of the list. The main differences are that $ does not allow computed indices, whereas [[ does, and that partial matching is allowed by default with operator $ but not with [[.

3. These Extract or Replacement operators come from S, although R restricts the use of partial matching, while S uses partial matching in most operators by default.

In your example, if CXCL1 and CXCL11 coexist and you index foo$CXCL1, that's not a partial match and should return CXCL1's value. If not, maybe there's another problem.

I should point out, [[ not allowing partial matching as default starts from version R 2.7.0 onwards.