[R-SIG-Mac] Problems in install some packages in R 4.2

[peter dalgaard]

We might want to circulate this in R Core, but offhand, I'd say that since the package ecosystem is somewhat platform dependent, it could be in the platform maintainers' domain to set defaults like

 options(pkgType="binary")

in the default installation's start-up files. This would make sense on Mac & Windows, but probably not on Linux variants. Users can always override it if needed.

(There's also an issue with pure R packages, which can be source-installed if the source is newer, but the byte-compile may still take a while and not be worth it for an incremental update.)

- Peter

> On 11 May 2022, at 07:56 , Simon Urbanek <simon.urbanek using r-project.org> wrote:
> 
> Peter,
> 
> I agree that the current behavior is not very helpful for the majority of users. Personally, I'd change the default "no" so that users that don't know or care what it entails can still get the package installed. Note that having a compiler may not be enough, since there are often library dependencies, so compiling from sources should really be attempted only by developers that do it deliberately. The only argument I see for using the "both" default is for cases where the package binary doesn't exist, but that is another story as the user has no choice there.
> 
> Cheers,
> Simon
> 
> PS: version mismatches are in principle expected due to package sources being released before binaries, but should be generally rare. However, there is currently an unusual back-log of package builds for R 4.2.0 on the high-sierra x86_64 system which should hopefully clear by tomorrow.
> 
> 
>> On 10/05/2022, at 9:18 PM, peter dalgaard <pdalgd using gmail.com> wrote:
>> 
>> Marginally related:
>> 
>> The "binary" bit trips up my students from time to time (with tseries recently). 
>> 
>> They get the option to install from source because it's marginally newer, answer "yes" because that'w what they always do, and then installation bombs because they don't have the compilers.  
>> 
>> Yes, they should know better, but no... 
>> 
>> We might want to consider implementing something like capabilities("compilers") so that we can condition the source installs more carefully.
>> 
>> 
>> -pd
>> 
>> 
>> 
>>> On 10 May 2022, at 04:47 , Simon Urbanek <simon.urbanek using r-project.org> wrote:
>>> 
>>> José,
>>> 
>>> can you be more specific as of what you did and exact output? I would recommend using a simple
>>> 
>>> install.packages("ggplot2",type="binary")
>>> 
>>> in R.
>>> 
>>> Cheers,
>>> Simon
>>> 
>>> 
>>>> On 10/05/2022, at 12:30 PM, José Carlos Guerrero Antúnez <jcgantunez using gmail.com> wrote:
>>>> 
>>>> Hi,
>>>> 
>>>> When I upgraded to R version 4.2 (Platform: x86_64-apple-darwin17.0
>>>> (64-bit)), I was having problems installing certain packages (e.g. ggplot2,
>>>> Hmisc, devtools) and their dependencies.
>>>> 
>>>> I got this error message "installation of package ‘package’ had non-zero
>>>> exit status.
>>>> 
>>>> I have tried several ways to install (e.g. Package Archive file (.tgz).
>>>> 
>>>> Any ideas on how to solve this problem, thank you very much.
>>>> 
>>>> 
>>>> -- 
>>>> José Carlos Guerrero
>>>> 
>>>> 	[[alternative HTML version deleted]]
>>>> 
>>>> _______________________________________________
>>>> R-SIG-Mac mailing list
>>>> R-SIG-Mac using r-project.org
>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>>>> 
>>> 
>>> _______________________________________________
>>> R-SIG-Mac mailing list
>>> R-SIG-Mac using r-project.org
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>> 
>> -- 
>> Peter Dalgaard, Professor,
>> Center for Statistics, Copenhagen Business School
>> Solbjerg Plads 3, 2000 Frederiksberg, Denmark
>> Phone: (+45)38153501
>> Office: A 4.23
>> Email: pd.mes using cbs.dk  Priv: PDalgd using gmail.com
>> 
> 

-- 
Peter Dalgaard, Professor,
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Office: A 4.23
Email: pd.mes using cbs.dk  Priv: PDalgd using gmail.com