Hi - I am trying to create buffers for large (~1GB) line and polygon vector files. I tried v.buffer but after letting it run overnight with no apparent progress I killed the process. I'm not aware of any software capable of buffering large vector files on a modest computer so that wasn't a big surprise. I then converted the vector to a large (23GB) raster file and am now trying r.buffer. I ran some tests on small images but can't figure out how the distance parameter works when an image has latitude/longitude coordinates. My image is in a geographic (latitude/longitude) projection with a resolution of 0.00025 (~27.75m) and I'd like to create a buffer with a distance of 0.0018 (~200m). How to I do that without projecting my image? I'm using GRASS 7.0 on 64 bit Ubuntu 10.04.
The manual seems to indicate that I can choose the distance units so I tried distances=200.0 and units=meters but that converted the entire image to a buffer so the distance was clearly too large. I then tried distances =0.0018 (the manual states that r.buffer uses "ellipsoidal geodesic distance measure for latitude/longitude locations) and units=meters but that didn't create any buffer. My last test was to use the number of pixels to define the distance so I tried distances =7.2 and units=meters and that created a buffer roughly one pixel wide.
Any thoughts on how to buffer a large vector or raster data set would be appreciated.
On Fri, Mar 25, 2011 at 2:42 PM, Ned Horning <horning@amnh.org> wrote:
Hi - I am trying to create buffers for large (~1GB) line and polygon vector
files. I tried v.buffer but after letting it run overnight with no apparent
progress I killed the process. I'm not aware of any software capable of
buffering large vector files on a modest computer so that wasn't a big
surprise.
So we hope that there is still a chance to make GRASS the only one
to suppor this
Note that v.buffer is somewhat in transition: http://trac.osgeo.org/grass/ticket/699#comment:10
"All these test cases are successfully completed by v.buffer/v.parallel;
on confirmation by I would reactivate v.buffer/v.parallel and deactivate
v.buffer2/v.parallel. For trunk that would mean that v.buffer which is
actually v.buffer2 gets replaced with v.buffer, same for v.parallel.
If you can compile yourself, please try GRASS 6.5 from current SVN
(or weekly snapshot) and go into
vector/v.buffer/
and compile this fixed older but apparently better version.
If that then works ok for you, it will be another reason to reactivate
the fixed original version of v.buffer rather than promoting the
currently used v.buffer2.
Thanks for your help. I am trying to follow your instruction but ran into a problem compiling v.buffer. I used svn to download grass 6.5 into /usr/local/src and it appears to have compiled just fine and grass65 seems to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that didn't work. I pasted the error message below.
I expect my MODULE_TOPDIR assignment is not correct. I'll try to figure out the problem tomorrow but if you know what the problem is please let me know.
Once I compile v.buffer do I have to do anything to deactivate v.buffer2?
Thanks again for the help.
Ned
--
nedhorning@ubuntu-laptop:/usr/local/src/grass6_devel/vector/v.buffer$ sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
/usr/local/grass-6.5.svn//include/Make/Module.make:25: warning: overriding commands for target `install'
/usr/local/grass-6.5.svn//include/Make/Rules.make:90: warning: ignoring old commands for target `install'
gcc -I/usr/local/grass-6.5.svn/dist.x86_64-unknown-linux-gnu/include -g -O2 -I/usr/include/gdal -DPACKAGE=\""grassmods"\" -I/usr/local/grass-6.5.svn/dist.x86_64-unknown-linux-gnu/include -o OBJ.x86_64-unknown-linux-gnu/main.o -c main.c
main.c: In function ‘main’:
main.c:274: error: ‘dist’ undeclared (first use in this function)
main.c:274: error: (Each undeclared identifier is reported only once
main.c:274: error: for each function it appears in.)
main.c:324: error: invalid type argument of ‘->’ (have ‘int’)
make: *** [OBJ.x86_64-unknown-linux-gnu/main.o] Error 1
On 03/26/2011 05:27 AM, Markus Neteler wrote:
Hi Ned,
On Fri, Mar 25, 2011 at 2:42 PM, Ned Horning<horning@amnh.org> wrote:
Hi - I am trying to create buffers for large (~1GB) line and polygon vector
files. I tried v.buffer but after letting it run overnight with no apparent
progress I killed the process. I'm not aware of any software capable of
buffering large vector files on a modest computer so that wasn't a big
surprise.
So we hope that there is still a chance to make GRASS the only one
to suppor this
Note that v.buffer is somewhat in transition: http://trac.osgeo.org/grass/ticket/699#comment:10
"All these test cases are successfully completed by v.buffer/v.parallel;
on confirmation by I would reactivate v.buffer/v.parallel and deactivate
v.buffer2/v.parallel. For trunk that would mean that v.buffer which is
actually v.buffer2 gets replaced with v.buffer, same for v.parallel.
If you can compile yourself, please try GRASS 6.5 from current SVN
(or weekly snapshot) and go into
vector/v.buffer/
and compile this fixed older but apparently better version.
If that then works ok for you, it will be another reason to reactivate
the fixed original version of v.buffer rather than promoting the
currently used v.buffer2.
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning <horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran into a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65 seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
On Sun, Mar 27, 2011 at 8:51 PM, Markus Neteler <neteler@osgeo.org> wrote:
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning <horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran into a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65 seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
Thanks for fixing the code. I managed to compile v.buffer and I ran it in the grass65 terminal window and these are the errors I got. There were several of the "Next edge was visited...." messages but I only cut out the last couple. Any idea what the problem might be? I'm running on 64 bit Ubuntu.
All the best,
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
*** glibc detected *** d.vect: free(): invalid next size (normal): 0x00000000013a5fc0 ***
======= Backtrace: =========
/lib/libc.so.6(+0x775b6)[0x7fc0f26965b6]
/lib/libc.so.6(cfree+0x73)[0x7fc0f269ce83]
/usr/local/grass-6.5.svn/lib/libgrass_vect.6.5.svn.so(Vect_destroy_line_struct+0x28)[0x7fc0f424f638]
d.vect(plot1+0xb0a)[0x40876a]
d.vect(main+0x1314)[0x4072d4]
/lib/libc.so.6(__libc_start_main+0xfd)[0x7fc0f263dc4d]
d.vect[0x403cd9]
======= Memory map: ========
00400000-0040c000 r-xp 00000000 08:05 206110 /usr/local/grass-6.5.svn/bin/d.vect
0060b000-0060c000 r--p 0000b000 08:05 206110 /usr/local/grass-6.5.svn/bin/d.vect
0060c000-0060d000 rw-p 0000c000 08:05 206110 /usr/local/grass-6.5.svn/bin/d.vect
01398000-013b9000 rw-p 00000000 00:00 0 [heap]
7fc0e4000000-7fc0e4021000 rw-p 00000000 00:00 0
7fc0e4021000-7fc0e8000000 ---p 00000000 00:00 0
7fc0e97cd000-7fc0e97d9000 r-xp 00000000 08:05 531120 /lib/libnss_files-2.11.1.so
7fc0e97d9000-7fc0e99d8000 ---p 0000c000 08:05 531120 /lib/libnss_files-2.11.1.so
7fc0e99d8000-7fc0e99d9000 r--p 0000b000 08:05 531120 /lib/libnss_files-2.11.1.so
7fc0e99d9000-7fc0e99da000 rw-p 0000c000 08:05 531120 /lib/libnss_files-2.11.1.so
7fc0e99da000-7fc0e99e4000 r-xp 00000000 08:05 531542 /lib/libnss_nis-2.11.1.so
7fc0e99e4000-7fc0e9be3000 ---p 0000a000 08:05 531542 /lib/libnss_nis-2.11.1.so
7fc0e9be3000-7fc0e9be4000 r--p 00009000 08:05 531542 /lib/libnss_nis-2.11.1.so
7fc0e9be4000-7fc0e9be5000 rw-p 0000a000 08:05 531542 /lib/libnss_nis-2.11.1.so
7fc0e9be5000-7fc0e9bed000 r-xp 00000000 08:05 527981 /lib/libnss_compat-2.11.1.so
7fc0e9bed000-7fc0e9dec000 ---p 00008000 08:05 527981 /lib/libnss_compat-2.11.1.so
7fc0e9dec000-7fc0e9ded000 r--p 00007000 08:05 527981 /lib/libnss_compat-2.11.1.so
7fc0e9ded000-7fc0e9dee000 rw-p 00008000 08:05 527981 /lib/libnss_compat-2.11.1.so
7fc0e9dee000-7fc0e9df1000 r-xp 00000000 08:05 523508 /lib/libgpg-error.so.0.4.0
7fc0e9df1000-7fc0e9ff0000 ---p 00003000 08:05 523508 /lib/libgpg-error.so.0.4.0
7fc0e9ff0000-7fc0e9ff1000 r--p 00002000 08:05 523508 /lib/libgpg-error.so.0.4.0
7fc0e9ff1000-7fc0e9ff2000 rw-p 00003000 08:05 523508 /lib/libgpg-error.so.0.4.0
7fc0e9ff2000-7fc0ea002000 r-xp 00000000 08:05 134688 /usr/lib/libtasn1.so.3.1.7
7fc0ea002000-7fc0ea201000 ---p 00010000 08:05 134688 /usr/lib/libtasn1.so.3.1.7
7fc0ea201000-7fc0ea202000 r--p 0000f000 08:05 134688 /usr/lib/libtasn1.so.3.1.7
7fc0ea202000-7fc0ea203000 rw-p 00010000 08:05 134688 /usr/lib/libtasn1.so.3.1.7
7fc0ea203000-7fc0ea21c000 r-xp 00000000 08:05 134634 /usr/lib/libsasl2.so.2.0.23
7fc0ea21c000-7fc0ea41b000 ---p 00019000 08:05 134634 /usr/lib/libsasl2.so.2.0.23
7fc0ea41b000-7fc0ea41c000 r--p 00018000 08:05 134634 /usr/lib/libsasl2.so.2.0.23
7fc0ea41c000-7fc0ea41d000 rw-p 00019000 08:05 134634 /usr/lib/libsasl2.so.2.0.23
7fc0ea41d000-7fc0ea433000 r-xp 00000000 08:05 531562 /lib/libresolv-2.11.1.so
7fc0ea433000-7fc0ea632000 ---p 00016000 08:05 531562 /lib/libresolv-2.11.1.so
7fc0ea632000-7fc0ea633000 r--p 00015000 08:05 531562 /lib/libresolv-2.11.1.so
7fc0ea633000-7fc0ea634000 rw-p 00016000 08:05 531562 /lib/libresolv-2.11.1.so
7fc0ea634000-7fc0ea636000 rw-p 00000000 00:00 0
7fc0ea636000-7fc0ea638000 r-xp 00000000 08:05 523515 /lib/libkeyutils-1.2.so
7fc0ea638000-7fc0ea837000 ---p 00002000 08:05 523515 /lib/libkeyutils-1.2.so
7fc0ea837000-7fc0ea838000 r--p 00001000 08:05 523515 /lib/libkeyutils-1.2.so
7fc0ea838000-7fc0ea839000 rw-p 00002000 08:05 523515 /lib/libkeyutils-1.2.so
7fc0ea839000-7fc0ea840000 r-xp 00000000 08:05 134268 /usr/lib/libkrb5support.so.0.1
7fc0ea840000-7fc0eaa3f000 ---p 00007000 08:05 134268 /usr/lib/libkrb5support.so.0.1
7fc0eaa3f000-7fc0eaa40000 r--p 00006000 08:05 134268 /usr/lib/libkrb5support.so.0.1
7fc0eaa40000-7fc0eaa41000 rw-p 00007000 08:05 134268 /usr/lib/libkrb5support.so.0.1
7fc0eaa41000-7fc0eaa65000 r-xp 00000000 08:05 132675 /usr/lib/libk5crypto.so.3.1
7fc0eaa65000-7fc0eac65000 ---p 00024000 08:05 132675 /usr/lib/libk5crypto.so.3.1
7fc0eac65000-7fc0eac66000 r--p 00024000 08:05 132675 /usr/lib/libk5crypto.so.3.1
7fc0eac66000-7fc0eac67000 rw-p 00025000 08:05 132675 /usr/lib/libk5crypto.so.3.1
7fc0eac67000-7fc0eac7e000 r-xp 00000000 08:05 527976 /lib/libnsl-2.11.1.so
7fc0eac7e000-7fc0eae7d000 ---p 00017000 08:05 527976 /lib/libnsl-2.11.1.so
7fc0eae7d000-7fc0eae7e000 r--p 00016000 08:05 527976 /lib/libnsl-2.11.1.so
7fc0eae7e000-7fc0eae7f000 rw-p 00017000 08:05 527976 /lib/libnsl-2.11.1.so
7fc0eae7f000-7fc0eae81000 rw-p 00000000 00:00 0
7fc0eae81000-7fc0eaef6000 r-xp 00000000 08:05 523504 /lib/libgcrypt.so.11.5.2
7fc0eaef6000-7fc0eb0f5000 ---p 00075000 08:05 523504 /lib/libgcrypt.so.11.5.2
7fc0eb0f5000-7fc0eb0f6000 r--p 00074000 08:05 523504 /lib/libgcrypt.so.11.5.2
7fc0eb0f6000-7fc0eb0f9000 rw-p 00075000 08:05 523504 /lib/libgcrypt.so.11.5.2
7fc0eb0f9000-7fc0eb195000 r-xp 00000000 08:05 134182 /usr/lib/libgnutls.so.26.14.12
7fc0eb195000-7fc0eb394000 ---p 0009c000 08:05 134182 /usr/lib/libgnutls.so.26.14.12
7fc0eb394000-7fc0eb39a000 r--p 0009b000 08:05 134182 /usr/lib/libgnutls.so.26.14.12
7fc0eb39a000-7fc0eb39b000 rw-p 000a1000 08:05 134182 /usr/lib/libgnutls.so.26.14.12
7fc0eb39b000-7fc0eb3a8000 r-xp 00000000 08:05 132150 /usr/lib/liblber-2.4.so.2.5.4 100%
Writting buffers...
100%
Building topology for vector map <testBuf2>...
Registering primitives...
1435962 primitives registered
115977416 vertices registered
Number of nodes: 1435838
Number of primitives: 1435962
Number of points: 0
Number of lines: 0
Number of boundaries: 1435962
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
GRASS 6.5.svn (NHD_Test):~ >
On 03/27/2011 02:51 PM, Markus Neteler wrote:
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning<horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran into a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65 seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
On Mon, Mar 28, 2011 at 12:39 PM, Ned Horning <horning@amnh.org> wrote:
Markus,
Thanks for fixing the code. I managed to compile v.buffer and I ran it in
the grass65 terminal window and these are the errors I got. There were
several of the "Next edge was visited...." messages but I only cut out the
last couple. Any idea what the problem might be? I'm running on 64 bit
Ubuntu.
All the best,
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
It seems that this is still v.buffer2, not v.buffer (a bit confusing
because the stuff in vector/v.buffer2 is compiled as v.buffer, as is
the stuff in vector/v.buffer). You need to either start grass from the
source tree or run make install again in order to make use of the
vector/v.buffer module.
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning<horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran into
a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65
seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
Thanks for the pointer but I'm still having a problem (terminal output pasted below) and wonder if I'm still accessing v.buffer2. The program stops because I run out of memory. I have 5.7 GB of physical and 5.8 GB for my swap and it's not enough. I ran make and make install in the v.buffer directory and ran configure, make and make install in /usr/local/src/grass6_devel. I expect I'm doing something stupid but am at a loss as to how to fix it.
Any thoughts?
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
100%
Writting buffers...
100%
Building topology for vector map <testBuf>...
Registering primitives...
1435962 primitives registered
115977416 vertices registered
Number of nodes: 1435838
Number of primitives: 1435962
Number of points: 0
Number of lines: 0
Number of boundaries: 1435962
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
On 03/28/2011 07:44 AM, Markus Metz wrote:
On Mon, Mar 28, 2011 at 12:39 PM, Ned Horning<horning@amnh.org> wrote:
Markus,
Thanks for fixing the code. I managed to compile v.buffer and I ran it in
the grass65 terminal window and these are the errors I got. There were
several of the "Next edge was visited...." messages but I only cut out the
last couple. Any idea what the problem might be? I'm running on 64 bit
Ubuntu.
All the best,
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
It seems that this is still v.buffer2, not v.buffer (a bit confusing
because the stuff in vector/v.buffer2 is compiled as v.buffer, as is
the stuff in vector/v.buffer). You need to either start grass from the
source tree or run make install again in order to make use of the
vector/v.buffer module.
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning<horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran into
a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65
seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
On Mon, Mar 28, 2011 at 11:40 PM, Ned Horning <horning@amnh.org> wrote:
Hi Markus M.
Thanks for the pointer but I'm still having a problem (terminal output
pasted below) and wonder if I'm still accessing v.buffer2. The program stops
because I run out of memory. I have 5.7 GB of physical and 5.8 GB for my
swap and it's not enough. I ran make and make install in the v.buffer
directory and ran configure, make and make install in
/usr/local/src/grass6_devel. I expect I'm doing something stupid but am at a
loss as to how to fix it.
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in /usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
Any thoughts?
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
100%
Writting buffers...
100%
Building topology for vector map <testBuf>...
Registering primitives...
1435962 primitives registered
115977416 vertices registered
Number of nodes: 1435838
Number of primitives: 1435962
Number of points: 0
Number of lines: 0
Number of boundaries: 1435962
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
On 03/28/2011 07:44 AM, Markus Metz wrote:
On Mon, Mar 28, 2011 at 12:39 PM, Ned Horning<horning@amnh.org> wrote:
Markus,
Thanks for fixing the code. I managed to compile v.buffer and I ran it in
the grass65 terminal window and these are the errors I got. There were
several of the "Next edge was visited...." messages but I only cut out
the
last couple. Any idea what the problem might be? I'm running on 64 bit
Ubuntu.
All the best,
Ned
--
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
WARNING: Next edge was visited but it is not the first one !!! breaking
loop
It seems that this is still v.buffer2, not v.buffer (a bit confusing
because the stuff in vector/v.buffer2 is compiled as v.buffer, as is
the stuff in vector/v.buffer). You need to either start grass from the
source tree or run make install again in order to make use of the
vector/v.buffer module.
On Sun, Mar 27, 2011 at 3:57 PM, Ned Horning<horning@amnh.org> wrote:
Hi Markus,
Thanks for your help. I am trying to follow your instruction but ran
into
a
problem compiling v.buffer. I used svn to download grass 6.5 into
/usr/local/src and it appears to have compiled just fine and grass65
seems
to be working. To compile v.buffer I tried:
sudo make MODULE_TOPDIR=/usr/local/grass-6.5.svn/
from the /usr/local/src/grass6_devel/vector/v.buffer directory but that
didn't work. I pasted the error message below.
I have fixed to typo style bugs. Now it compiles, please try again.
On Tue, Mar 29, 2011 at 8:27 AM, Markus Metz
<markus.metz.giswork@googlemail.com> wrote:
...
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in /usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
An alternative: simply modify
vector/Makefile
to point to v.buffer and not v.buffer2...
Thanks for the help compiling with v.buffer. I think I'm running the correct version now but the function still does not complete properly. I pasted the last several lines from the terminal screen below. I have a feeling I ran out of memory. Any ideas about what I can try next?
Ned
--
D0/0: sa: 22
D0/0: current: 22
D0/0: current = 29, last = 30, lcount = 1
D0/0: sa: 28
D0/0: current: 28
D0/0: current = 29, last = 34, lcount = 1
D0/0: sa: 28
D0/0: current: 28
Building parts of topology...
Building topology for vector map <testBuf>...
Registering primitives...
1433691 primitives registered
119731336 vertices registered
Number of nodes: 1433605
Number of primitives: 1433691
Number of points: 0
Number of lines: 0
Number of boundaries: 1433691
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
GRASS 6.5.svn (NHD_Test):~ >
On Tue, Mar 29, 2011 at 8:27 AM, Markus Metz
<markus.metz.giswork@googlemail.com> wrote:
...
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in /usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
An alternative: simply modify
vector/Makefile
to point to v.buffer and not v.buffer2...
On Wed, Mar 30, 2011 at 1:52 PM, Ned Horning <horning@amnh.org> wrote:
Thanks for the help compiling with v.buffer. I think I'm running the correct
version now but the function still does not complete properly. I pasted the
last several lines from the terminal screen below. I have a feeling I ran
out of memory. Any ideas about what I can try next?
Not sure if the "d.vect: free(): invalid" error means that d.vect
caused the problem.
You could try several things:
- update your grass-6.5 to r45873
- close the grass gui and other applications not really necessary
- now try v.buffer again
The process got killed while snapping boundaries which is rather
memory demanding. Memory consumption for vector operations like
snapping can be substantially reduced in grass 7, but grass 7 still
has only the broken version of v.buffer2 (to be fixed).
Alternatively you could use r.grow.distance on the rasterized version
and set all values larger than the desired buffer distance to NULL
Markus M
--
D0/0: sa: 22
D0/0: current: 22
D0/0: current = 29, last = 30, lcount = 1
D0/0: sa: 28
D0/0: current: 28
D0/0: current = 29, last = 34, lcount = 1
D0/0: sa: 28
D0/0: current: 28
Building parts of topology...
Building topology for vector map <testBuf>...
Registering primitives...
1433691 primitives registered
119731336 vertices registered
Number of nodes: 1433605
Number of primitives: 1433691
Number of points: 0
Number of lines: 0
Number of boundaries: 1433691
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
GRASS 6.5.svn (NHD_Test):~ >
On Tue, Mar 29, 2011 at 8:27 AM, Markus Metz
<markus.metz.giswork@googlemail.com> wrote:
...
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in
/usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
An alternative: simply modify
vector/Makefile
to point to v.buffer and not v.buffer2...
Thanks once again for the advice. I updated to the most recent version yesterday and tried v.buffer again but it ran out of memory at around the 75% complete mark during the snapping boundaries step. I only have 5.7 GB physical memory and 5.7 GB swap available. I tried r.grow and that appears to have worked just fine. It took about 10 hours but I'm not in a rush so that's fine. I look forward to testing the GRASS 7 version of v.buffer when that is working.
All the best,
Ned
On 04/03/2011 09:07 AM, Markus Metz wrote:
On Wed, Mar 30, 2011 at 1:52 PM, Ned Horning<horning@amnh.org> wrote:
Thanks for the help compiling with v.buffer. I think I'm running the correct
version now but the function still does not complete properly. I pasted the
last several lines from the terminal screen below. I have a feeling I ran
out of memory. Any ideas about what I can try next?
Not sure if the "d.vect: free(): invalid" error means that d.vect
caused the problem.
You could try several things:
- update your grass-6.5 to r45873
- close the grass gui and other applications not really necessary
- now try v.buffer again
The process got killed while snapping boundaries which is rather
memory demanding. Memory consumption for vector operations like
snapping can be substantially reduced in grass 7, but grass 7 still
has only the broken version of v.buffer2 (to be fixed).
Alternatively you could use r.grow.distance on the rasterized version
and set all values larger than the desired buffer distance to NULL
Markus M
--
D0/0: sa: 22
D0/0: current: 22
D0/0: current = 29, last = 30, lcount = 1
D0/0: sa: 28
D0/0: current: 28
D0/0: current = 29, last = 34, lcount = 1
D0/0: sa: 28
D0/0: current: 28
Building parts of topology...
Building topology for vector map<testBuf>...
Registering primitives...
1433691 primitives registered
119731336 vertices registered
Number of nodes: 1433605
Number of primitives: 1433691
Number of points: 0
Number of lines: 0
Number of boundaries: 1433691
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
GRASS 6.5.svn (NHD_Test):~>
On Tue, Mar 29, 2011 at 8:27 AM, Markus Metz
<markus.metz.giswork@googlemail.com> wrote:
...
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in
/usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
An alternative: simply modify
vector/Makefile
to point to v.buffer and not v.buffer2...
On Tue, Apr 5, 2011 at 2:52 PM, Ned Horning <horning@amnh.org> wrote:
Markus M.
Thanks once again for the advice. I updated to the most recent version
yesterday and tried v.buffer again but it ran out of memory at around the
75% complete mark during the snapping boundaries step. I only have 5.7 GB
physical memory and 5.7 GB swap available. I tried r.grow and that appears
to have worked just fine. It took about 10 hours but I'm not in a rush so
that's fine. I look forward to testing the GRASS 7 version of v.buffer when
that is working.
The GRASS 7 version of v.buffer is now working.
Some preprocessing you can try before buffering:
- clean the vector with v.clean tool=break,rmdupl
- simplify the vector with v.generalize using as threshold buffer
distance * tolerance, in this case 0.0018 * 0.01
To avoid out-of-memory errors, first export GRASS_VECTOR_LOWMEM=1 [1],
buffer lines and areas separately, then patch and clean the two
results
1. v.buffer type=line distance=0.0018 output=buffer1
2. v.buffer type=area distance=0.0018 output=buffer2
3. v.patch input=buffer1,buffer2 output=buffer
4. v.clean input=buffer output=buffer_clean tool=bpol,rmdupl,break,rmdupl
some more cleaning may be necessary
Please let me know if it works, and if not, where it fails.
On Wed, Mar 30, 2011 at 1:52 PM, Ned Horning<horning@amnh.org> wrote:
Thanks for the help compiling with v.buffer. I think I'm running the
correct
version now but the function still does not complete properly. I pasted
the
last several lines from the terminal screen below. I have a feeling I ran
out of memory. Any ideas about what I can try next?
Not sure if the "d.vect: free(): invalid" error means that d.vect
caused the problem.
You could try several things:
- update your grass-6.5 to r45873
- close the grass gui and other applications not really necessary
- now try v.buffer again
The process got killed while snapping boundaries which is rather
memory demanding. Memory consumption for vector operations like
snapping can be substantially reduced in grass 7, but grass 7 still
has only the broken version of v.buffer2 (to be fixed).
Alternatively you could use r.grow.distance on the rasterized version
and set all values larger than the desired buffer distance to NULL
Markus M
--
D0/0: sa: 22
D0/0: current: 22
D0/0: current = 29, last = 30, lcount = 1
D0/0: sa: 28
D0/0: current: 28
D0/0: current = 29, last = 34, lcount = 1
D0/0: sa: 28
D0/0: current: 28
Building parts of topology...
Building topology for vector map<testBuf>...
Registering primitives...
1433691 primitives registered
119731336 vertices registered
Number of nodes: 1433605
Number of primitives: 1433691
Number of points: 0
Number of lines: 0
Number of boundaries: 1433691
Number of centroids: 0
Number of areas: -
Number of isles: -
Snapping boundaries...
Killed
GRASS 6.5.svn (NHD_Test):~>
On Tue, Mar 29, 2011 at 8:27 AM, Markus Metz
<markus.metz.giswork@googlemail.com> wrote:
...
If you ran make in the v.buffer directory and afterwards configure,
make and make install in /usr/local/src/grass6_devel, you would end up
with v.buffer2.
1. run configure, make, but not make install in
/usr/local/src/grass6_devel
2. run make in the v.buffer directory, watch out for errors to make
sure it compiles all right
3. run make install in /usr/local/src/grass6_devel, but not again
configure and/or make
4. that should give you the original v.buffer
An alternative: simply modify
vector/Makefile
to point to v.buffer and not v.buffer2...
On Fri, Mar 25, 2011 at 2:42 PM, Ned Horning <horning@amnh.org> wrote:
Hi - I am trying to create buffers for large (~1GB) line and polygon vector
files. I tried v.buffer but after letting it run overnight with no apparent
progress I killed the process. I'm not aware of any software capable of
buffering large vector files on a modest computer so that wasn't a big
surprise. I then converted the vector to a large (23GB) raster file and am
now trying r.buffer. I ran some tests on small images but can't figure out
how the distance parameter works when an image has latitude/longitude
coordinates.
For the record, Markus Metz fixed r.buffer in GRAS 7 for latlong
(r50792).
