[R-sig-Geo] Linux or OS X on Macbook Pro Retina?

[Roger Bivand]

On Wed, 19 Jun 2013, Rainer M Krug wrote:

> Roger Bivand <Roger.Bivand at nhh.no> writes:
>
>> On Wed, 19 Jun 2013, Rainer M Krug wrote:
>>
>>>
>>> Barry Rowlingson <b.rowlingson at lancaster.ac.uk> writes:
>>>
>>>> On Wed, Jun 19, 2013 at 9:13 AM, Rainer M Krug <Rainer at krugs.de> wrote:
>>>>> Hi
>>>>>
>>>>> I finally received my new computer - a Mac Powerbook Retina.
>>>
>>> Sorry - Macbook Pro Retina
>>>
>>>>>
>>>>> I am a Linux / Ubuntu user, but I am thinking about possibly using OS
>>>>> X. How does OS X play with these spatial applicatios, mainly R (and
>>>>> associated packages), gdal, proj4, GRASS? Are there problems /
>>>>> complications with using spgrass6 under OS X?
>>>>>
>>>>> Any experiences, things to be wary about?
>>>>
>>>>  Go back about two messages on this mailing list and I asked a similar
>>>> question, only without a Mac to play on!
>>>
>>> True - I have seen that one. But apart from these, I am looking for some
>>> more insight from actual users of OS X.
>>
>> Although I'm not an actual user, I think that it will be important to
>> install GRASS and its dependencies first, then the R packages using
>> those components (rgdal, rgeos, etc) from source, linking dynamically
>> to the underlying dylibs, as one would typically do on Linux. One
>> reason for this is to maintain consistency in the PROJ and GDAL
>> metadata (things like EPSG), which differ for different versions of
>> say PROJ.
>
> That makes sense - so to install e.g. rgdal, I would use
>
> ,----
> | install.packages("rgdal, type="source")
> `----

You will need to specify the configure arguments, I think - rather as in 
Barry's description, or in file.show(system.file("README", 
package="rgdal")) - to avoid recursion, on a machine with rgdal already 
installed! Or:

https://r-forge.r-project.org/scm/viewvc.php/pkg/inst/README?view=markup&root=rgdal

>
> after before installing the dependencies. I will try it out by
> installing the dependencies vie Macports, which makes it quite
> easy.

The Kyngchaos frameworks are also well-regarded, and also provide an rgdal 
binary package built against the frameworks.

Roger

>
> Cheers,
>
> Rainer
>
>
>> Both GRASS and R developers often use OSX as Unix, but they
>> expect to build more components from source where sensible.
>>
>> Roger
>>
>>>
>>>>
>>>>  Roger's answer is that for most people you can just do
>>>> install.packages and get binary versions of those packages with static
>>>> linking so all dependencies are resolved and there's no need to go
>>>> hunting for source code and development packages.
>>>
>>> That is most definitely very useful.
>>>
>>>>
>>>>  If you can try that and it all works, please let me know - I'm trying
>>>> to write some instructions for new users for a workshop, and I guess
>>>> it will be 60% windows, 35% mac, 5% linux...
>>>
>>> OK - I will see.
>>>
>>> Thanks,
>>>
>>> Rainer
>>>
>>>>
>>>> Barry
>>>
>>>
>>>
> <#secure method=pgpmime mode=sign>
>
>

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no