[GRASS-user] [R-sig-Geo] [R] Need help using GRASS within R - error when running the script for a second time
Roger Bivand
Roger.Bivand at nhh.no
Thu Jul 9 06:24:36 PDT 2020
Dear Loïc,
On Wed, 8 Jul 2020, Loïc Valéry wrote:
> Dear Roger,
>
> Following the message you left on github this morning (here is the github link to make it easier for the subscribers to follow our exchanges: https://github.com/rsbivand/rgrass7/issues/13#issuecomment-655384113), I continued my investigations and I finally found the origin of the problem.
>
> First of all, I would like to answer your questions this morning:
> - OS: Windows 8.1
> - R version 3.6.3
> - all the packages are up to date
> - the GRASS version: stand-alone Windows GRASS 7.8.3
> - I confirm that when I load the library rgrass7, R returns a message indicating that GRASS is not open.
>
> Coming back to the subject itself, when you told me that you were unable
> to reproduce the problem under Windows, I also did the test and, indeed,
> when I only ran these few lines of code, everything worked fine and I
> could execute initGRASS() as many times as I wanted without any
> problems.
>
> But these few lines of code are included in a larger script that first
> calls link2GI to interface R with the OTB software. And this is where
> the problem occurs. Here are the results of the tests in a very
> synthetic way:
So perhaps raise the issue with link2GI?
https://github.com/r-spatial/link2GI/
Or separate the components to run in different R sessions as a stop-gap?
It is quite likely that attaching link2GI will overwrite GRASS environment
variables (and the reverse is probably the case too).
Roger
>
> Test 1:
> 1-Running initGRASS() n times: everything works fine n times
> 2-Running link2GI::linkOTB()
> 3-Running initGRASS() : this crashes on the first attempt.
>
> Test 2:
> 1-Running link2GI::linkOTB()
> 2-Running initGRASS() n times: everything works fine n times
> 3-Running link2GI::linkOTB()
> 4-Running initGRASS() : this crashes on the first attempt.
>
> There is clearly a conflict between these two packages: running the command link2GI::linkOTB() seems to "break" the link made by rgrass7 between R and GRASS.
>
> So, I modified the code to prevent the problem from occurring during a second run of the script. But even though I had no problems since then, it's still a "band-aid on a wooden leg". It would be nice if Chris could find a solution to prevent linkOTB() from breaking the link between R and GRASS.
>
> I hope this feedback is clear and remain at your disposal to complete this message if you would like more information.
>
> Best regards,
> Loïc
>
>
>
> De : R-sig-Geo <r-sig-geo-bounces at r-project.org> de la part de Loïc Valéry <lvalery at outlook.fr>
> Envoyé : mardi 7 juillet 2020 21:03
> À : Roger.Bivand at nhh.no <Roger.Bivand at nhh.no>
> Cc : grass-user <grass-user at lists.osgeo.org>; r-sig-geo at r-project.org <r-sig-geo at r-project.org>
> Objet : Re: [R-sig-Geo] [R] Need help using GRASS within R - error when running the script for a second time
>
> Dear Roger,
>
> Sorry for cross-posting.
> As you invite me, I'm going to post my case on https://github.com/rsbivand/rgrass7/issues.
>
> Just a short comment to yours: if the reason you give is the right one, I don't understand why the script works the first time but not the second time. In your hypothesis, I guess it shouldn't work at all.
>
> Best regards,
> Loïc
>
> De : Roger Bivand <Roger.Bivand at nhh.no>
> Envoyé : mardi 7 juillet 2020 20:20
> À : Loïc Valéry <lvalery at outlook.fr>
> Cc : Micha Silver <tsvibar at gmail.com>; r-sig-geo at r-project.org <r-sig-geo at r-project.org>; grass-user <grass-user at lists.osgeo.org>
> Objet : Re: [R-sig-Geo] [R] Need help using GRASS within R - error when running the script for a second time
>
> Please do not cross-post between lists, because few are likely to
> subscribe to both. Choose rather to post only to the list likely to
> include people using R and GRASS:
>
> http://lists.osgeo.org/mailman/listinfo/grass-stats
>
> Further, please post plain text only - I missed your earlier post to
> grass-user because it was scrubbed as HTML (or maybe Micha's reply).
>
> Once resolved, you might report back on grass-user and R-sig-geo on the
> findings.
>
> Briefly, it looks as though you are caught in the transition in sp/rgdal
> and sf from Proj4 to WKT2-2019 CRS, and similar changes in Grass. Grass
> 7.8 and R packages may be installed with old PROJ/GDAL or new PROJ/GDAL.
> So far, nobody knows how to use g.proj and siblings sensibly in the R
> context. Most likely, using the wkt= parameter when the object in R has a
> WKT2 representation is going to be much more secure than passing a proj4=
> string, but wkt= takes a file name or standard input.
>
> I would invite discussion in an issue on:
>
> https://github.com/rsbivand/rgrass7/issues
>
> with use cases (platform-independent).
>
> Roger
>
> On Tue, 7 Jul 2020, Loïc Valéry wrote:
>
>> Dear all,
>>
>> As advised by Micha (FYI his reply is at the end of this email) , I
>> contact you to submit an inextricable problem when using rgrass7 : the
>> script (cf. just below) works the first time but, when I rerun it a
>> second time, R and windows return error messages. For this to work, I
>> must close and open R or use the rs.restartR() command from RStudio,
>> which is not very convenient because it stops the script.
>>
>> Many thanks in advance for your help.
>> Loïc
>
>> @Micha : thank you very much for your reply. But I performed the tests
>> you asked me to do and, unfortunately, none of them solve the problem.
>> Also, I tried to run "execGRASS("g.proj", flags = "c", proj4 = p4str)" a
>> second time WITHOUT reinitializing GRASS but R and windows still return
>> the same error messages.
>>
>> ERRORS WHEN I RUN THE SCRIPT FOR A SECOND TIME :
>>
>> - FROM WINDOWS : g.region.exe : the procedure entry point GEOSMakeValid_r could not be located in the dynamic link library C:\Program Files\GRASS GIS 7.8\extrabin\gdal300.dll
>>
>> - FROM R :
>> # Link to GRASS GIS software v.7.8.3
>>> initGRASS(gisBase ="C:/Program Files/GRASS GIS 7.8",
>> + home="temp/GRASS_tmp", use_g.dirseps.exe=F,
>> + gisDbase="temp/GRASS_tmp", mapset="PERMANENT",
>> + remove_GISRC=T, override=T)
>> Error in if (is.na(projstr)) uprojargs <- projstr else uprojargs <- paste(unique(unlist(strsplit(projstr, :
>> l'argument est de longueur nulle
>> De plus : There were 50 or more warnings (use warnings() to see the first 50)
>>>
>>> # Specifying the projection reference for the GRASS working environment
>>> p4str<-sp::proj4string(seg_poly)
>> Warning message:
>> In sp::proj4string(seg_poly) :
>> CRS object has comment, which is lost in output
>>>
>>> execGRASS("g.proj", flags = "c", proj4 = p4str)
>>> rgrass7::use_sp()
>>>
>>> # Converting the 'sp' object into a GRASS-readable file format (here, 'vec1.shp')
>>> writeVECT(seg_poly,"vec1",v.in.ogr_flags=c("o", "overwrite"),
>> + driver="ESRI Shapefile")
>> Error in writeVECT(seg_poly, "vec1", v.in.ogr_flags = c("o", "overwrite"), :
>> driver %in% candDrivers is not TRUE
>> De plus : Warning message:
>> In system(syscmd, intern = intern, ignore.stderr = ignore.stderr, :
>> l'exécution de la commande 'v.in.ogr.exe -f' renvoie un statut 313
>>>
>>>
>>> # Run the 'v.generalize' function
>>> execGRASS("v.generalize",flag=c("overwrite"),
>> + parameters=list(input="vec1",
>> + output="GRASS_smooth_seg_poly",
>> + error="GRASS_smooth_seg_poly_error",
>> + method="distance_weighting",
>> + threshold=1))
>>>
>>>
>>> # Converting the shapefile 'GRASS_smooth_seg_poly.shp' into a R-readable object
>>> # (here, a SpatialPolygonDataFrame named 'smooth_seg_poly')
>>> smooth_seg_poly<-readVECT("GRASS_smooth_seg_poly", with_prj=T,
>> + driver="ESRI Shapefile")
>> Error in .read_vect_non_plugin(vname = vname, layer = layer, type = type, :
>> driver %in% candDrivers is not TRUE
>> De plus : Warning message:
>> In system(syscmd, intern = intern, ignore.stderr = ignore.stderr, :
>> l'exécution de la commande 'v.in.ogr.exe -f' renvoie un statut 313
>>
>>
>>
>> REPLY FROM MICHA :
>> De : Micha Silver <tsvibar at gmail.com>
>> Envoyé : mardi 7 juillet 2020 15:55
>> À : Loïc Valéry <lvalery at outlook.fr>
>> Cc : grass-user <grass-user at lists.osgeo.org>
>> Objet : Re: [R] Need help using GRASS within R - error when running the script for a second time
>>
>> Hello Loïc
>>
>> Let's keep the discussion on the list, so others can help and benefit from responses.
>>
>> BTW, you might want to post to the r-sig-geo list also. If you come across questions specific to R and spatial functions, that's the best place to ask.
>>
>> Below,inline, some guesses as to what is happening...
>>
>> On 07/07/2020 0:03, Loïc Valéry wrote:
>> initGRASS(gisBase ="C:/Program Files/GRASS GIS 7.8",
>>
>> + home="temp/GRASS_tmp", use_g.dirseps.exe=F,
>> + gisDbase="temp/GRASS_tmp", mapset="PERMANENT",
>> + remove_GISRC=T, override=T)
>> Error in if (is.na(projstr)) uprojargs <- projstr else uprojargs <- paste(unique(unlist(strsplit(projstr, :
>> l'argument est de longueur nulle
>> De plus : There were 50 or more warnings (use warnings() to see the first 50)
>>
>> I'm not sure, but I don't think you want to rerun initGRASS() a second time. In your script, you do initGRASS only once, then if you need to perform several analyses, do them in the same (running) GRASS session.
>>
>>
>> # Specifying the projection reference for the GRASS working environment
>> p4str<-sp::proj4string(seg_poly)
>>
>> Warning message:
>> In sp::proj4string(seg_poly) :
>> CRS object has comment, which is lost in output
>>
>> execGRASS("g.proj", flags = "c", proj4 = p4str)
>> rgrass7::use_sp()
>>
>> # Converting the 'sp' object into a GRASS-readable file format (here, 'vec1.shp')
>> writeVECT(seg_poly,"vec1",v.in.ogr_flags=c("o", "overwrite"),
>>
>> + driver="ESRI Shapefile")
>>
>> The driver name is "ESRI_Shapefile" (note the underscore). Perhaps that is the problem here?
>>
>> Error in writeVECT(seg_poly, "vec1", v.in.ogr_flags = c("o", "overwrite"), :
>> driver %in% candDrivers is not TRUE
>> De plus : Warning message:
>> In system(syscmd, intern = intern, ignore.stderr = ignore.stderr, :
>> l'exécution de la commande 'v.in.ogr.exe -f' renvoie un statut 313
>>
>> # Run the 'v.generalize' function
>> execGRASS("v.generalize",flag=c("overwrite"),
>>
>> + parameters=list(input="vec1",
>> + output="GRASS_smooth_seg_poly",
>> + error="GRASS_smooth_seg_poly_error",
>> + method="distance_weighting",
>> + threshold=1))
>>
>> # Converting the shapefile 'GRASS_smooth_seg_poly.shp' into a R-readable object
>> # (here, a SpatialPolygonDataFrame named 'smooth_seg_poly')
>> smooth_seg_poly<-readVECT("GRASS_smooth_seg_poly", with_prj=T,
>>
>> + driver="ESRI Shapefile")
>>
>> Same here, the driver name for shapefile is wrong.
>>
>> I would also add that both R and GRASS have switched to the better Geopackage format for exchanging spatial data. You might want to try, the driver name is "GPKG"
>>
>> See details about Geopackage here:
>> https://www.gis-blog.com/geopackage-vs-shapefile/
>> and in French here:
>> https://www.sigterritoires.fr/index.php/le-format-geopackage-et-qgis-3/
>>
>> Best, Micha
>>
>>
>>
>>
>>
>>
>> PREVIOUS EMAILS:
>>
>> De : Micha Silver <tsvibar at gmail.com>
>> Envoyé : mardi 9 juin 2020 23:08
>> À : Loïc Valéry <lvalery at outlook.fr>
>> Objet : Re: [R] Need help using GRASS within R - problem with CRS using the 'v.generalize' command
>>
>>
>> On 09/06/2020 16:22, Loïc Valéry wrote:
>> Dear Micha,
>>
>> A thousand thanks for your help. It works now!
>> Whoa, you'd better save some thanks for the next question. ;-)
>>
>> For your information and following your remark, I also updated rgdal.
>> Thanks again for taking the time to answer me.
>> Kind regards
>> Loïc
>>
>> De : Micha Silver <tsvibar at gmail.com>
>> Envoyé : mardi 9 juin 2020 08:44
>> À : Loïc Valéry <lvalery at outlook.fr>; r-help at R-project.org <r-help at R-project.org>
>> Objet : Re: [R] Need help using GRASS within R - problem with CRS using the 'v.generalize' command
>>
>> Hello
>>
>>
>> On 08/06/2020 19:52, Loïc Valéry wrote:
>>
>> Dear all,
>>
>> First of all, this is my first message on the list. Therefore, please be indulgent if my message is not perfectly formatted as it should be.
>>
>> I am currently encountering a difficulty with GRASS 7.8 within R when using the 'v.generalize' command to smooth the contour of polygons after a segmentation step.
>>
>> I tried two different ways to "call" GRASS:
>>
>> 1 - using the RQGIS3 package
>>
>> setwd("D:/test")
>> library(sp)
>> library(rgdal)
>>
>> rgdal: version: 1.4-8, (SVN revision 845)
>> Geospatial Data Abstraction Library extensions to R successfully loaded
>> Loaded GDAL runtime: GDAL 2.2.3, released 2017/11/20
>> Path to GDAL shared files: C:/Users/toto/Documents/R/win-library/3.6/rgdal/gdal
>> GDAL binary built with GEOS: TRUE
>> Loaded PROJ.4 runtime: Rel. 4.9.3, 15 August 2016, [PJ_VERSION: 493]
>> Path to PROJ.4 shared files: C:/Users/toto/Documents/R/win-library/3.6/rgdal/proj
>> Linking to sp version: 1.4-1
>>
>> Please note that you have older versions of GDAL (2.2.3 here) and PROJ.4
>> (4.9 here). These are currently being replaced by GDAL 3.0 and PROJ 6.3.
>> You might (should) want to follow the discussion the the r-sig-geo maillist:
>>
>>
>> https://stat.ethz.ch/pipermail/r-sig-geo/2020-June/028165.html
>>
>>
>> ....... (skipped all the discussion regarding RQGIS3 as I think it's not
>> relevant)
>>
>> setwd("D:/test")
>> library(rgrass7)
>>
>> # characteristics of the SpatialPolygonsDataFrame 'seg_poly' : CRS does exist
>> seg_poly
>>
>> class : SpatialPolygonsDataFrame
>> features : 31
>> extent : 477371.3, 477397.6, 5631995, 5632020 (xmin, xmax, ymin, ymax)
>> crs : +proj=utm +zone=32 +datum=WGS84 +units=m +no_defs
>> variables : 1
>> names : Seg_ID
>> min values : 1
>> max values : 31
>> Warning message:
>> In proj4string(x) : CRS object has comment, which is lost in output
>>
>> # initialization of GRASS 7.8 from R
>> initGRASS(gisBase ="C:/Program Files/GRASS GIS 7.8", home="temp/GRASS",gisDbase="temp/GRASS", use_g.dirseps.exe=F,remove_GISRC=T, override=T)
>>
>> gisdbase temp/GRASS
>> location file19685026c56
>> mapset file196829fa7141
>> rows 1
>> columns 1
>> north 1
>> south 0
>> west 0
>> east 1
>> nsres 1
>> ewres 1
>> projection NA
>>
>> I suspect that the problem comes from 'projection NA' when initializing GRASS (cf. just above)
>>
>> What you need to do here is setup the CRS of your new location.
>> Typically, you would run initGRASS and point to a *previously created*
>> LOCATION, with CRS already defined. In this case, since you are creating
>> a new location, you must define it's coordinate system. (GRASS is very
>> "picky" about that).
>>
>> Here's a GIS Stackexchange post that explains:
>>
>> https://gis.stackexchange.com/questions/183032/create-a-new-grass-database-in-r-with-crs-projection
>>
>>
>> You can derive the full proj4 string from your sp object with:
>>
>>
>> p4str = sp::proj4string(seg_poly)
>>
>>
>> Then use that to set the project parameters for the new LOCATION, with
>>
>> execGRASS("g.proj", flags = "c", proj4 = p4str) Now you should be able
>> to continue with...
>>
>>
>>
>> execGRASS("v.generalize",flag=c("overwrite"),parameters=list(input="vec1",
>>
>> + output="GRASS_smooth_seg_poly",
>> + error="GRASS_smooth_seg_poly_error",
>> + method="distance_weighting",
>> + threshold=1))
>>
>>
>>
>> As a novice in the field, I would be very grateful for your help.
>> I remain at your entire disposal for any further information you may need to help me in finding a solution to this problem.
>>
>> Yours sincerely,
>> Loïc
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>>
>>
>
> --
> Roger Bivand
> Department of Economics, Norwegian School of Economics,
> Helleveien 30, N-5045 Bergen, Norway.
> voice: +47 55 95 93 55; e-mail: Roger.Bivand at nhh.no
> https://orcid.org/0000-0003-2392-6140
> https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
--
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; e-mail: Roger.Bivand at nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
More information about the grass-user
mailing list