[R-sig-Geo] suggestion for spgrass6 - temp file in readVECT6

[Roger Bivand]

On Wed, 15 Dec 2010, Pierre Roudier wrote:

> Hi Roger,
>
> Thanks for the patch. Unfortunately it does not work on my box, but I
> suspect that to be related with my readVECT6() problem.
>
> foo <- readVECT6('bar', driver='GML')
> Error in ogrInfo(dsn = dsn, layer = layer, input_field_name_encoding =
> input_field_name_encoding) :
>  Cannot open layer
>
> writeVECT6(foo, 'deletethis', driver='GML')
> Available OGR Drivers:
> Error in writeOGR(SDF, dsn = rtmpfl1, layer = shname, driver = driver) :
>
>        GDAL Error 4: Failed to create GML file
> /home/roudierp/Documents/GRASSDATA/NZMG/bh/.tmp/A208_RoudierP.
>
> So I guess I'd better fix that problem first>
>
> Sorry I can't provide better testing for now,

This thread continues on the grass-stats list. The driver= argument to the 
interface is not a good idea, as the dsn/olayer strings needed by the 
drivers differ greatly, and providing for many combinations is not a 
robust solution. Pierre's problems are compounded by trying to transfer an 
enormous data set, which may also challenge the file size limit for 
shapes, among other things. Using the GML driver is also problematic, as 
it is using WKT text, creating an even larger file.

Roger

>
> Cheers,
>
> Pierre
>
> 2010/12/15 Roger Bivand <Roger.Bivand at nhh.no>:
>> On Fri, 10 Dec 2010, Roger Bivand wrote:
>>
>>> On Thu, 9 Dec 2010, Dylan Beaudette wrote:
>>>
>>>> On Thursday, December 09, 2010, Pierre Roudier wrote:
>>>>>
>>>>> Dear all,
>>>>>
>>>>> I got a suggestion for spgrass6. As far as I understand it, a
>>>>> temporary ESRI shapefile is written on the disk when using
>>>>> readVECT6('foo'), I guess then R is using rgdal to read that shapefile
>>>>> back in memory. Considering the various limitations of the ESRI
>>>>> shapefile format that does not occur with the GRASS internal format,
>>>>> could it be possible to add an option to choose the OGR format used in
>>>>> that process?
>>>>>
>>>>> Also got a question for Roger I guess: I'm using now the grass7 svn
>>>>> trunk, should I expect any problems using the spgrass6 interface (I
>>>>> mean problems due to structural differences that appears in grass7
>>>>> that would not be handled by spgrass6, not problems due to grass7
>>>>> itself).
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Pierre
>>>>>
>>>>
>>>> This is an excellent point Pierre, and something that has caused
>>>> significant pain the past-- in terms of column names and NULL value
>>>> handling. Not everyone's GDAL/OGR will be configured in the same way-- so I
>>>> don't know if altering the default behavior would be wise. However, adding
>>>> an option (as you suggested) to use some other format, such as SQLite, would
>>>> give power users a nice alternative to SHP format. SQLite would allow for
>>>> longer field names, proper support for NULL data, and probably some other
>>>> 'good things'.
>>>
>>> SQLite is an extra OGR driver that will be dependent on the OGR being used
>>> on the platform in question. The OGR used by GRASS may be the same as that
>>> used by rgdal, but may not be (R binary rgdal packages for Windows and OSX,
>>> which have minimal driver sets).
>>>
>>> I'll attempt to add a driver= argument to readVECT6() and writeVECT6(),
>>> and drivername= readRAST6() and writeRAST6() for the temporary files.
>>
>> Could either of you, or others who may be interested, please try the draft
>> version by anonymous checkout fot the spgrass6 module from the sourceforge
>> r-spatial project, or using the source package at:
>>
>> http://spatial.nhh.no/R/Devel/spgrass6_0.6-24.tar.gz
>>
>> If you report back that it suits your needs, I'll submit it to CRAN; if you
>> need further changes, please let me know.
>>
>> Roger
>>
>>>
>>> Typically, for reading native GRASS data, users build GRASS plugins for
>>> OGR/GDAL. readVECT6() and readRAST6() both by default check for the presence
>>> of the plugin, and use it if present, avoiding any intermediate files being
>>> created.
>>>
>>> The recently released revision of spgrass6 (0.6-23) does work with GRASS >
>>> 6.3, including GRASS 7. This is signalled by the use of "GRASS 6+" rather
>>> than just "GRASS 6" in its title and short description.
>>>
>>> Hope this helps,
>>>
>>> Roger
>>>
>>>>
>>>> Dylan
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>> --
>> Roger Bivand
>> Economic Geography Section, Department of Economics, Norwegian School of
>> Economics and Business Administration, 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
>>
>> _______________________________________________
>> R-sig-Geo mailing list
>> R-sig-Geo at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>
>

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, 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