[Rd] Rscript and default packages
Duncan Murdoch
murdoch at stats.uwo.ca
Fri Sep 4 16:42:48 CEST 2009
On 9/4/2009 10:20 AM, Simon Urbanek wrote:
> On Sep 3, 2009, at 13:52 , Romain Francois wrote:
>
>> On 09/03/2009 05:23 PM, Duncan Murdoch wrote:
>>>
>>> On 03/09/2009 9:53 AM, Romain Francois wrote:
>>>> Hi,
>>>>
>>>> Is is possible to embed inside an R script, the name of the default
>>>> packages to be loaded when the script is invoked with Rscript.
>>>>
>>>> I know about the --default-packages argument, but I was wondering if
>>>> there was a mechanism to embed this information within the script
>>>> itself
>>>
>>> I don't understand what you'd want here that you don't get with
>>> attach()
>>> or require(). Why does it matter if they are default?
>>>
>>> Duncan Murdoch
>>
>> Sorry for being vague. I am more interested in not loading some
>> packages:
>>
>> $ time Rscript -e "#"
>>
>> real 0m0.224s
>> user 0m0.188s
>> sys 0m0.032s
>>
>>
>> $ time Rscript --default-packages="base" -e "#"
>>
>> real 0m0.067s
>> user 0m0.033s
>> sys 0m0.016s
>>
>>
>> $ time r -e "#"
>>
>> real 0m0.039s
>> user 0m0.032s
>> sys 0m0.006s
>>
>> This is related to the "How to ship R scripts with R packages"
>> thread. I'd like for example to ship a script that I know only
>> requires the "base" package. How would I specify this from within
>> the script.
>>
>
> Well, what's wrong with:
>
> #!/usr/bin/Rscript --default-packages=base
>
> ginaz:sandbox$ time ./scr
>
> real 0m0.045s
> user 0m0.027s
> sys 0m0.017s
>
> .. as opposed to
>
> #!/usr/bin/Rscript --default-packages=base
>
> ginaz:sandbox$ time ./scr
>
> real 0m0.201s
> user 0m0.166s
> sys 0m0.034s
Windows doesn't support the #! invocation, but that should work on all
the other systems that would use scripts.
Duncan Murdoch
More information about the R-devel
mailing list