[R] About source()

[Li SUN]

Thanks, Rolf, Justin and Uwe!

Actually I wanted to run .R file as a script, just like what people do
for bash scripts or python scripts. It seems to me that adding
"#!/usr/bin/R -f" at the first line is what I need. Is this true?

Li Sun


2012/1/2 Rolf Turner <rolf.turner at xtra.co.nz>:
> On 03/01/12 17:02, Li SUN wrote:
>>
>> Hello,
>>
>> I am a beginner to the R language and find it fantastic and
>> well-designed, quite different from other programming languages.
>
>    What a refreshingly sensible attitude!!! :-)
>
>> This is the first time I post on the r-help mailing list.
>>
>> In invoking the function source(filename), it seems that the filename
>> has to exist in the current working directory, otherwise it has to be
>> specified in full path. So is there any mechanism(such as environment
>> variable) to specify an additional directory of .R files that source()
>> could search in?
>
>
> I'm no expert on this, and others may correct me, but
>
>    (1) I don't believe any such mechanism exists.
>
>    (2) It's probably not a good idea, even if such a mechanism
>    were to exist.  Directories have a tree structure, rather than
>    being linearly ordered in the way that data bases on your
>    R search path are ordered.
>
>    I believe you would run all sorts of risks of confusion and
>    of getting the wrong file were you to invoke such a mechanism.
>
>    It is ``good practice'' to have separate directories associated
>    with different projects and to situate all files, that you might wish
>    to source in respect of a given project, in the directory associated
>    with that project.
>
> I know that this is an irritating sort of response --- ``No, you can't
> do that, and you shouldn't do it anyway!'' --- but I sincerely believe
> this to be true.
>
> That being said, I also believe that you could program up such a mechanism
> yourself. I.e. build a function "source2()" which would have a "hard coded"
> list
> of directories to search, and would make use of the "try()" function.
>
> Might be a good exercise for you, given that you are starting out in R
> and looking to upgrade your skills! :-)
>
>    cheers,
>
>        Rolf Turner
>
>