[GRASS-dev] Direct link from GRASS to R via C

[Rainer M Krug]

Vaclav Petras <wenzeslaus at gmail.com> writes:

> On 16 July 2013 09:59, Rainer M Krug <Rainer at krugs.de> wrote:
>>
>> Vaclav Petras <wenzeslaus at gmail.com> writes:
>>
>>> Hi,
>> Hi,
>>
>>>
>>> you probably know, I've once started the discussion on this topic on
>>> grass-stats ML (very simple test code included) [1]. And I've also
>>> tried some basic example and experienced some problems [2].
>>
>> To my shame, I must say I overlooked your threads. I am aware of the
>> problem which Roger discussed, therefore I think it would be the best to
>> write a function in GRASS which is exported and is doing the access and
>> simply returns the column / block / complete raster (i.e. a *very basic*
>> exported raster access api for grass).
>>
>> This function could be called then from Rcpp and it would be much easier
>> then writing the complete access function in Rcpp (as you mention below,
>> I also think that the header files locations might be a problem).
>>
>>
>>> To the
>>> information you provided I would add that the new functionality have
>>> to be written in C++ since Rcpp is only for C++.
>>
>> I am always confused by these two and how they are linked together - can
>> I use C libraries from Rcpp, as described above?
>>
> Yes you can, see my example. Generally, you can call C functions from
> C++ code and create C wrapper for your C++ library (you can find
> documentation everywhere, note especially the extern "C" declaration).

Thanks - I'll look into it.

>
>> As you already have done some work (could you please mail it to me as I
>> can't access the attachment via gmane?), I would very much appreciate
>> your input - also my C / C++ experience is quite limited and input from
>> that side would be brilliant.
>>
> I have only that small example (attached) but it contains all you
> need, however there were also some problems [2].

Thanks - that looks interesting (and easier then I thought, judging from
the r.example included in GRASS). I will look at it in detail and come
back to you if I have any questions.

Thanks,

Rainer

>
>> Thanks,
>>
>> Rainer
>>
>>>
>>> Vaclav
>>>
>>> [1] http://lists.osgeo.org/pipermail/grass-stats/2012-July/001422.html
>>> [2] http://lists.r-forge.r-project.org/pipermail/rcpp-devel/2012-July/004055.html
>>>
>>>
>>> On 15 July 2013 17:30, Rainer M Krug <Rainer at krugs.de> wrote:
>>>> The following message is a courtesy copy of an article
>>>> that has been posted to gmane.comp.gis.grass.devel as well.
>>>>
>>>> Hi
>>>>
>>>> I posted on my blog an outline of an idea of how spatial data could be
>>>> easily loaded from GRASS into R and written back.
>>>>
>>>> (see http://rmkrug.wordpress.com/2013/07/15/grassrlink-1/)
>>>>
>>>> In a nutshell, the idea would be to use the C functions in GRASS to
>>>> write a function which returns a single column, column range or whole
>>>> raster. This function could then be called from R (via Rcpp) to get the
>>>> data into R without having to worry about intermittent exports and
>>>> imports via the hdd.
>>>>
>>>> I would ike to have some input from the GRASS developer community what
>>>> they think about such a function. This function should be usable as
>>>> "stand alone", and not requiring the opening and buffer allocations et
>>>> al as in the r.example, but rather simply take the mapset, raster name,
>>>> column(s) to read, if MASK should be respected, ... arguments.
>>>>
>>>> I was thinking that it would be useful to have such a function in GRASS,
>>>> as the compilation together with GRASS, in the same line as a module,
>>>> would be quite easy.
>>>>
>>>> Initially, I was thinking about read and write support for rasters,
>>>> which then could be extended to vectors and 3d-rasters and possibly even
>>>> the temporal data.
>>>>
>>>> Could you give feedback on what y9ou think about the idea, and how you
>>>> think it could be realized (simplicity to install would be important).
>>>>
>>>> Cheers and thanks,
>>>>
>>>> Rainer
>>>>
>>>> --
>>>> Rainer M. Krug
>>>>
>>>> email: RMKrug<at>gmail<dot>com
>>>> _______________________________________________
>>>> grass-dev mailing list
>>>> grass-dev at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/grass-dev
>>
>>
>> --
>> Rainer M. Krug
>>
>> email: RMKrug<at>gmail<dot>com
>>
>> _______________________________________________
>> grass-dev mailing list
>> grass-dev at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/grass-dev
>
> _______________________________________________
> grass-dev mailing list
> grass-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-dev
<#secure method=pgpmime mode=sign>

-- 
Rainer M. Krug

email: RMKrug<at>gmail<dot>com