[Rd] intersect() change of behavior in 4.2

[Martin Maechler]

>>>>> Lluís Revilla 
>>>>>     on Tue, 12 Apr 2022 09:16:36 +0200 writes:

    > Hi all,
    > This change is documented on the man page so I think it is intentional.

yes, also if you look at the (svn) log messages of the code changes
(or its git mirrors).

    >> From https://search.r-project.org/R/refmans/base/html/sets.html:

    > For union, a vector of a common mode.
    > For intersect, a vector of a common mode, or NULL if x or y is NULL.
    > For setdiff, a vector of the same mode as x.

    > Now the results are symmetrical to intersect( "foo", list())

indeed, and that *is* very desirable.

    > Probably it is worth mentioning in the NEWS as it was found to cause
    > some test to fail on a Bioconductor package some months ago.
    > This could affect other packages and analysis too.

    > Best,
    > Lluís

I agree and have added an entry there ... still to be
back ported to R 4.2.0 beta.

Martin


    > On Tue, 12 Apr 2022 at 01:22, Gábor Csárdi <csardi.gabor using gmail.com> wrote:
    >> 
    >> I wonder if this change is intentional, and if it is, whether it is
    >> worth mentioning in the NEWS.
    >> 
    >> ❯ R-4.1 -q -e 'intersect(list(), "foo")'
    >> > intersect(list(), "foo")
    >> character(0)
    >> 
    >> ❯ R-4.2 -q -e 'intersect(list(), "foo")'
    >> > intersect(list(), "foo")
    >> list()
    >> 
    >> ______________________________________________
    >> R-devel using r-project.org mailing list
    >> https://stat.ethz.ch/mailman/listinfo/r-devel

    > ______________________________________________
    > R-devel using r-project.org mailing list
    > https://stat.ethz.ch/mailman/listinfo/r-devel