[R] Message produced under R 3.6.0.

Martin Maechler m@ech|er @end|ng |rom @t@t@m@th@ethz@ch
Tue Apr 30 11:31:51 CEST 2019

>>>>> Duncan Murdoch 
>>>>>     on Mon, 29 Apr 2019 21:54:47 -0400 writes:

    > On 29/04/2019 9:44 p.m., Rolf Turner wrote:
    >> On 30/04/19 1:31 PM, Duncan Murdoch wrote:
    >>> On 29/04/2019 9:25 p.m., Duncan Murdoch wrote:
    >>>> On 29/04/2019 8:43 p.m., Rolf Turner wrote:
    >>>>> On 30/04/19 12:19 PM, Jeff Newmiller wrote:
>>>>> Since you now have this indirect dependency, you should make sure you
>>>>> have updated this gaggle of packages. I have found that the
>>>>> dependencies do not necessarily update when I update the package that
>>>>> relies on them. It can take a few passes  of reading error/warming
>>>>> messages to get them all updated. I suspect you have old versions of
>>>>> packages that now delegate certain functions to lower-level packages,
>>>>> and these errors should go away when you have them all current.
    >>>>> Nope.  Doesn't work.  I reinstalled brms and the messages still appeared
    >>>>> when I loaded brms.
    >>>> Those messages probably aren't due to brms directly: for example, the
    >>>> first set
    >>>>   > Registered S3 methods overwritten by 'ggplot2':
    >>>>   >   method         from
    >>>>   >   [.quosures     rlang
    >>>>   >   c.quosures     rlang
    >>>>   >   print.quosures rlang
    >>>> is due to issues with ggplot2 or rlang.  Have you tried updating both of
    >>>> those?
    >>> To answer my own question:  this one still appears after updating those
    >>> two, but it appears to be a bug in ggplot2, because that package
    >>> replaces those functions in a file compat-quosures.R with comment
    >>> # TODO: Remove once rlang or later is on CRAN
    >>> and the rlang version on CRAN is now 0.3.4.
    >>> Duncan Murdoch
    >> Thanks Duncan.  So I guess I should just wait until Hadley gets around
    >> to fixing ggplot2?

    > I think so.

    >> OTOH --- I would, on general principles, like to learn how to suppress
    >> such messages ....

    > I'm not sure you (as package author) can, but your users can do it via

    > suppressMessages(library(brms))

    > (or presumably the same wrapper when attaching your package, or some 
    > other package that loads ggplot2 or xts).  It appears that dplyr has 
    > already been fixed.

    > Duncan Murdoch

After all, I think the new messages have shown their worth here:

They were only "uttered" because of bugs
(.. and maybe the bugs have only be fixed so quickly because of
 these messages appearing?)

I don't think you (or others) should suppress such messages by
default.  They do indicate someone should do something, even if
it's not you.

Somewhat related, note that the NEWS for 3.6.0 also contain [not prominently
  enough maybe: this is a quite nice set of features which
  notably package maintainers can / should start making use of !] :

    • library() and require() now allow more control over handling
      search path conflicts when packages are attached. The policy is
      controlled by the new conflicts.policy option.


More information about the R-help mailing list