On Sun, 29 Aug 2010, William McCoy wrote:

Since this issue was first brought up on this list and then continued on grass-stats and grass-dev, I think I should summarize what has been found out so far regarding this issue.

As mentioned below, I have grass63 installed on this computer along with the grass64svn that I was trying to access via initGRASS() in spgrass6. Apparently, the grass63 libraries were being accessed before the grass64svn libraries were found and, hence, the symbol lookup error.

I have found that if I modify initGRASS() to prepend the grass64svn directories, rather than append them, to PATH and LD_LIBRARY_PATH, the error does not occur. I don't know if this is the best solution, but it works temporarily and at least we know the source of the problem. See here for more details on the source of the problem and the temporary "fix":

http://lists.osgeo.org/pipermail/grass-stats/2010-August/001268.html

Thanks to Roger Bivand, Markus Neteler, and Glynn Clements (in chronological order) for sorting this out.


I will most likely adopt this ordering in the next release of spgrass6, as it then matches the ordeing that GRASS itself uses on startup. When there are multiple installations of interfaced software components, it can happen.

Roger


Bill


On 08/28/2010 02:09 PM, William McCoy wrote:
On 08/28/2010 01:57 PM, Roger Bivand wrote:
On Sat, 28 Aug 2010, William McCoy wrote:

On 08/28/2010 01:15 PM, Roger Bivand wrote:
On Sat, 28 Aug 2010, William McCoy wrote:

Roger,

I think your suspicions are correct. Here is the output from debug
after the PATH setting if():

Browse[2]> Sys.getenv("PATH")



PATH
"/usr/local/texlive/2009/bin/i386-linux:/usr/lib/qt-3.3/bin:/usr/kerberos/sbin:/usr/kerberos/bin:/usr/lib/ccache:/usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/wdmccoy/bin:/usr/local/grass-6.4.0svn/bin:/usr/local/grass-6.4.0svn/scripts"


Browse[2]> list.files(Sys.getenv("PATH"))
character(0)
Browse[2]> Sys.getenv("LD_LIBRARY_PATH")


LD_LIBRARY_PATH
"/usr/lib/R/lib:/usr/local/lib:/usr/lib/jvm/jre/lib/i386/server:/usr/lib/jvm/jre/lib/i386:/usr/lib/jvm/java/lib/i386:/usr/java/packages/lib/i386:/lib:/usr/lib:/usr/local/grass-6.4.0svn/lib"


Browse[2]> list.files(Sys.getenv("LD_LIBRARY_PATH"))
character(0)


Since I last used spgrass6 many months ago, I have installed
grass64svn along with grass63, which was long ago installed as a
fedora rpm. So my PATH looks like this:


[wdmc...@boreas ~]$ echo $PATH
/usr/local/texlive/2009/bin/i386-linux:/usr/lib/qt-3.3/bin:/usr/kerberos/sbin:/usr/kerberos/bin:/usr/lib/ccache:/usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/wdmccoy/bin





But my LD_LIBRARY_PATH looks like this:

[wdmc...@boreas ~]$ echo $LD_LIBRARY_PATH


I.e., there is nothing in LD_LIBRARY_PATH. The reason grass64 works by
itself (outside of R) is because I have a grass64.conf file in
/etc/ld.so.conf.d and that .conf file contains:

/usr/local/grass-6.4.0svn/lib


What do you recommend?

This is a bit difficult to say, but trying to set LD_LIBRARY_PATH
before
starting R to the correct value:

export LD_LIBRARY_PATH=/usr/local/grass-6.4.0svn/lib

and/or ensuring that ldconfig was run, might help.

Have you tried using gisBase = "/usr/local/grass-6.4.0svn" without the
terminating "/"?

Hope this goes in the right direction,

Roger


Yes, I had already tried gisBase = "/usr/local/grass-6.4.0svn" without
the terminating "/". In fact, that's what I had used when I did the
debugging on initGRASS(). So that doesn't seem to help.

I think ldconfig is always run on startup, but I ran it again just now
and it made no difference. And setting and exporting LD_LIBRARY_PATH
doesn't help either.

I should also mention that I had tried to use my grass63 installation
for gisBase a couple of times and that gave me the same results.

In case I didn't mention it, this is all being done on Fedora 13. The
last time that I used spgrass6 successfully was on Fedora 12 on the
same computer.

Any other ideas?

Trying ldd on GRASS modules when running GRASS outside R to work out
which shared object library they are using. I run Fedora 13 (among
others) so I doubt that it is the system, rather some changes in your
installed software that occurred at around the same time. Using ldd may
help. When I type:

ldd /home/rsb/topics/grass/g64_rc6/grass-6.4.0RC6/bin/g.region

I see an extra directory level before the /bin, /lib, etc., both in the
input and the output, but your mileage may vary.

Roger

PS: Could I suggest moving this thread to statsgrass, which is
specifically for the R/GRASS interface? Could you summarise there to
attract attention from GRASS people (who are not too busy with FOSS4G in
Barcelona)?


I will try summarizing this issue in a message to statsgrass this
afternoon.

With regard to ldd, I don't see the extra directory level:

[wdmc...@boreas ~]$ ldd /usr/local/grass-6.4.0svn/bin/g.region
linux-gate.so.1 => (0x0036c000)
libgrass_vect.so => /usr/local/grass-6.4.0svn/lib/libgrass_vect.so
(0x00110000)
libgrass_dbmibase.so =>
/usr/local/grass-6.4.0svn/lib/libgrass_dbmibase.so (0x00f52000)
...


Thanks, Bill




Thanks, Bill




Thanks, Bill


On 08/28/2010 05:41 AM, Roger Bivand wrote:
On Fri, 27 Aug 2010, William McCoy wrote:

I am using spgrass6, but I can't seem to initiate a grass session. I
managed to do this correctly months ago, but can't seem to now. Here
is my code and sessionInfo:

Something has happened to your GRASS installation and/or the
LD_LIBRARY_PATH environment variable. The GRASS binary modules are
being
found, but the libraries (shared objects) they call are not. Try to
set
debug(initGRASS), then step through looking at what is in environment
variables, etc. After the if () for conditionally setting the PATH,
look
at:

Sys.getenv("PATH")
list.files(Sys.getenv("PATH"))

and similarly for LD_LIBRARY_PATH:

Sys.getenv("LD_LIBRARY_PATH")
list.files(Sys.getenv("LD_LIBRARY_PATH"))

I suspect that the test:

eLDPATH <- Sys.getenv("LD_LIBRARY_PATH")
grep(basename(Sys.getenv("GISBASE")), eLDPATH)

is not empty, so the LD_LIBRARY_PATH does not get edited to include
the
actual locations of the *.so files.

Have you changed the definitions of these environment variables in a
.bashrc or similar between this working and no longer working? Before
starting R, say:

echo $PATH
echo $LD_LIBRARY_PATH

to see if that sheds any light on things.

Hope this helps,

Roger



library(spgrass6)
Loading required package: sp
Loading required package: rgdal
Geospatial Data Abstraction Library extensions to R successfully
loaded
Loaded GDAL runtime: GDAL 1.7.2, released 2010/04/23
Path to GDAL shared files: /usr/local/share/gdal
Loaded PROJ.4 runtime: Rel. 4.7.1, 23 September 2009
Path to PROJ.4 shared files: (autodetected)
Loading required package: XML
GRASS GIS interface loaded with GRASS version: (GRASS not running)
loc <- initGRASS(gisBase = "/usr/local/grass-6.4.0svn/", home =
tempdir())
g.gisenv: symbol lookup error: g.gisenv: undefined symbol:
G__no_gisinit
g.gisenv: symbol lookup error: g.gisenv: undefined symbol:
G__no_gisinit
g.gisenv: symbol lookup error: g.gisenv: undefined symbol:
G__no_gisinit
g.gisenv: symbol lookup error: g.gisenv: undefined symbol:
G__no_gisinit
g.gisenv: symbol lookup error: g.gisenv: undefined symbol:
G__no_gisinit
g.region: symbol lookup error: g.region: undefined symbol:
G__gisinit
Error in if (file.exists(file) == FALSE) if (!missing(asText) &&
asText == :
argument is of length zero
Error in parseGRASS(cmd) : g.region not parsed

sessionInfo()
R version 2.11.1 (2010-05-31)
i386-redhat-linux-gnu

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=C LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] spgrass6_0.6-19 XML_3.1-1 rgdal_0.6-28 sp_0.9-66

loaded via a namespace (and not attached):
[1] grid_2.11.1 lattice_0.18-8 tools_2.11.1


Any ideas about what I am doing wrong?

Thanks, Bill

















--
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.biv...@nhh.no

_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo

Reply via email to