[R] Troubles installing Rcmdr on Mac

[John Fox]

Dear Eberhard,

On 2021-01-12 12:32 a.m., Dr Eberhard W Lisse wrote:
> John,
> 
> what is wrong with installing Xcode’s command lime tools (not Xcode itself)?

Nothing, and I did miss the distinction, but it shouldn't be necessary, 
and the instructions for installing the Rcmdr are already more 
complicated on macOS than on other platforms because of the necessity to 
install XQuartz. Users should be able to install the Rcmdr package on 
macOS without having to install packages from source.

Remember that Rcmdr users are typically students in basic statistics 
courses, many of whom have limited software sophistication. 
Unnecessarily complicating the installation is undesirable. Of course, 
if it's necessary to complicate the installation, one has to live with that.

I'll be interested to learn whether my suggestions solve the problem. If 
not, I can add an instruction concerning the Xcode tools to the Rcmdr 
installation notes for macOS.

Thanks for your help,
  John

> 
> 
> —
> Sent from Dr Lisse’s iPhone
> On 12 Jan 2021, 04:30 +0200, John Fox <jfox using mcmaster.ca>, wrote:
>> Dear Stephane,
>>
>> I've taken yet another look at this and have an additional suggestion
>> for your students to try:
>>
>> install.packages("Rcmdr", type="mac.binary")
>>
>> That should avoid any attempt to install Rcmdr package dependencies from
>> source.
>>
>> I hope this helps,
>> John
>>
>> On 2021-01-11 3:53 p.m., John Fox wrote:
>>> Dear Stephane and Eberhard,
>>>
>>> As an addendum to my previous response, I uninstalled the Rcmdr package
>>> and all of its direct and indirect dependencies and then reinstalled the
>>> package -- on a macOS 11.1 system running R 4.0.3 with all other
>>> packages up-to-date.
>>>
>>> I then reinstalled the Rcmdr and dependencies via the command
>>> install.packages("Rcmdr"), and responded "no" when asked whether to
>>> install some packages from source (perhaps this is the explanation for
>>> the problem, if your students responded "yes" without having Xcode
>>> installed).
>>>
>>> Following these steps, everything (still) works fine. I therefore can't
>>> duplicate your students' problem, which makes it hard to suggest how to
>>> fix it, without having some additional details.
>>>
>>> Best,
>>>   John
>>>
>>>
>>> On 2021-01-11 3:33 p.m., John Fox wrote:
>>>> Dear Stephane and Eberhard,
>>>>
>>>> It should not be necessary to install Xcode (which includes otools) to
>>>> install and use the Rcmdr package on macOS because it shouldn't be
>>>> necessary to install the CRAN packages required from source. I'm
>>>> currently running the Rcmdr on two macOS 11.1 systems, with all CRAN
>>>> packages up-to-date, and don't have any problems.
>>>>
>>>> Stephane, have you and your students checked the Rcmdr installation
>>>> notes (at
>>>> <https://socialsciences.mcmaster.ca/jfox/Misc/Rcmdr/installation-notes.html>)
>>>> and followed the instructions there? If you have, and still experience
>>>> this problem, it would help to have some more information about what
>>>> they did to install the Rcmdr and what happened.
>>>>
>>>> In the meantime, I'll try a fresh install of the Rcmdr and
>>>> dependencies to see whether I encounter any difficulties.
>>>>
>>>> Best,
>>>>    John
>>>>
>>>
>>> ______________________________________________
>>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>> PLEASE do read the posting guide
>>> http://www.R-project.org/posting-guide.html
>>> and provide commented, minimal, self-contained, reproducible code.
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>