[R-SIG-Mac] Cocoa gui with embedded R framework

[stefano iacus]

Kjell,
there are other issues that also Simon raised. Even if you don't move 
R.app while running, if you try to install packages, then some 
configuration parameters needs to be changed as well (the headers won't 
be in the right place); you cannot use CRAN binaries, etc. You need to 
patch different things and probably an external script could do the 
job...but you need to run it every time you move R.app (and no, the 
Finder allow you to move R while it is running!!!)

It depends on what the aim of all of this is: if, for example, the aim 
is to let non admin users to install R, then one solution could be to 
install the R.framework in user's ~/Library/Frameworks and use a 
post-install script in the installer to set the correct paths. This way 
you can move around R.app (it will look for the R.framework in user's 
home) and relocate (once) the R.framework (not only the .dylib but also 
the R  scripts).

stefano

On 23/mar/05, at 21:33, Kjell Konis wrote:

> Simon,
>
> The way I am imagining this working out is that the linked path and 
> R_HOME will both be set dynamically relative to the location of R.app 
> when it is launched.  The @executable_path token seems to do the trick 
> for the linked path.  R.app always finds libR.dylib in the embedded 
> framework even if the R_HOME variable is not set.  Tomorrow I will try 
> to get Rcontroller to set R_HOME relative to the location of R.app at 
> launch.  Then (again the way I am imagining things) everything should 
> work unless you move R.app while it is running - which hopefully the 
> OS won't let you do.
>
> Kjell
>
> On Mar 23, 2005, at 8:05 PM, Simon Urbanek wrote:
>
>>> In any case, I would be pleased to know if is works after moving 
>>> around R.app on the system or even after installing a package and 
>>> move again R.app around.
>>
>> It won't really solve the duality issue: linked path and R_HOME. 
>> Technically it's the same as the current Info.plist solution. Both 
>> allow relocation of R.app, but none of them allows relocation of the 
>> framework.
>>
>> Cheers,
>> Simon
>
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>