[R-SIG-Mac] install.packages default type change?
Dan Putler
putler at sauder.ubc.ca
Sat Jul 23 05:49:21 CEST 2005
The notion that "I doubt there are many R users on MacOS X who do not
have the Developer Tools installed" is a problem in classroom
settings, where you can fairly easily explain to students who use a
Mac that they need to load X11, but having them also load XCode
tools, and then have them deal with the g77/gfortran issues would
likely be beyond most of them.
On 22-Jul-05, at 8:38 PM, Bill Northcott wrote:
> On 22/07/2005, at 8:00 PM, Simon Urbanek wrote:
>
>>> In the past, I believe that MacOS X users could use
>>> install.packages()
>>> successfully without any package 'type' specification, even though
>>> there
>>> is no .tgz file, but now it seems that they need to explicitly say:
>>>
>>> install.packages(..., type="source")
>>>
>>> Right?
>>>
>>> Is this a permanent change?
>>>
>>>
>> yes
>>
>>
>>> Is this a permanent change?
>>>
>>>
>>
>> FWIW: strictly speaking this is not a new change - the documentation
>> was describing exactly this behavior for a while - it was merely a
>> bug that the package installation implementation wasn't really
>> following the documentation.
>>
>> The main reason for this behavior is that unlike other unix systems
>> OS X comes by default without development tools, so you simply cannot
>> use type="source". Therefore the default is to use binary packages
>> which is the recommended way.
>>
>
> I am not sure this is a good decision. MacOS X is always supplied
> with the Xcode development tools, which is not true of all the UNIXen
> I have dealt with. It is really only Linux that I know to install
> developer tools as part of the default install and even then it is
> usually necessary to install header packages to actually build
> anything. Finally I doubt that there are many R users on MacOS X who
> do not have the Developer Tools installed, after all, they need X11
> which is also an optional install. Indeed you supply g77 as part of
> the CRAN binary, and that is useless if you don't have the Developer
> Tools.
>
> Bill Northcott
>
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>
More information about the R-SIG-Mac
mailing list