Skip to content

cairoDevice and GTK under MacOS

3 messages · Simon Urbanek, John Maindonald

Original
John Maindonald
# 
Arising from my recent interchange with Professor Ripley,
it seems that there is no current binary for cairoDevice
that works under R 2.11.0 or 2.11.1.  This is in spite of the
current CRAN package check results that have OK for
cairoDevice under r-release MacOS X ix86

With each new release of R (and maybe release of OSX), 
there seems a sizeable risk that the cairoDevice/GTK 
combination will fail.

Is cairoDevice/GTK  in for the long haul?  Or is it likely
to be superseded by other GUI interface approaches?
If it is in for the long haul, what are the odds of getting
quickly to a point where one can have confidence that
it will be usable under OSX when a new version of R 
appears?  

This is not a criticism of the excellent work that has gone
into building cairoDevice/GTK.  Restriction of some
software to some systems has to be accepted as a fact
of life, especially where there is a large element of
innovation.  Rather, it is a request for any insight that 
may be available from those on this list who have some 
limited ability to peer into the future, or even actual
knowledge of the future!

I guess that some part of the answer will in due course
come from commercial users of R.   What is available
on the Mac may become more important now that Google
is, apparently, giving up support internally for use of 
Windows systems.
http://www.ft.com/cms/s/2/d2f3f04e-6ccf-11df-91c8-00144feab49a.html

John Maindonald             email: john.maindonald at anu.edu.au
phone : +61 2 (6125)3473    fax  : +61 2(6125)5549
Centre for Mathematics & Its Applications, Room 1194,
John Dedman Mathematical Sciences Building (Building 27)
Australian National University, Canberra ACT 0200.
http://www.maths.anu.edu.au/~johnm
Simon Urbanek
#
On Jun 5, 2010, at 12:03 AM, John Maindonald wrote:

            

      
      
      
    

        
Seems?!? Based on what?!? Brian said clearly that there is a binary and it works for him -- so it does for me!

You never told us what *you* are using so there is not much we can do for you - from the output it seemed that you are using the wrong packages with wrong R versions so this is not about CRAN at all. His comments were based on the assumption that you are trying to build something for yourself because the CRAN packages work fine.

Cheers,
Simon

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      John Maindonald
    

    

      
      #
    

  


  

      
I did explain that I had installed the latest binary.  I had also tried to update cairoDevice.

I found that I had to remove the 2.10 library, then reinstall cairoDevice, and all is well.  Presumably cairoDevice was not updating in the 2.11 library (because the apparently identical version was in the 2.10 library?)  Updating cairoDevice did not, on its own, seem to be enough.

I find it surprising that I have to explicitly remove the 2.10 library.  I presume that this is a quirk, and can be removed in later releases.

What I needed to know was that others who were using the binary had been able to install and use cairoDevice.

I was confused by Brian's comment that:

"
(E.g. R 2.12.0 will require API version 7, but cairoDevice will not currently compile under pre-2.12.0: if you try the version compiled under 2.11.x you do get this message.)
"

if you try the version compiled under 2.11.x [with R 2.12.0] you do get this message.

John Maindonald             email: john.maindonald at anu.edu.au
phone : +61 2 (6125)3473    fax  : +61 2(6125)5549
Centre for Mathematics & Its Applications, Room 1194,
John Dedman Mathematical Sciences Building (Building 27)
Australian National University, Canberra ACT 0200.
http://www.maths.anu.edu.au/~johnm

        
On 06/06/2010, at 12:01 AM, Simon Urbanek wrote: