v.buffer2: difference between Linux 32bit and 64bit

[mmetz]

v.buffer2 (not original v.buffer) produces warning messages on Linux 64bit that are not apparent on Linux 32bit. The initial results differ slightly, the final result seems to be ok after lots of cleaning.

Reproducible with roadsmajor in nc_spm:

GRASS> v.buffer2 input=roadsmajor@PERMANENT output=roadsmajor_buf1_2000 distance=1000

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

No such warnings, but topological errors with distance=500, warnings like these but no topological errors with distance=1000 or larger.

[neteler]

The current v.buffer is working well, closing.

[mmetz]

Reopening because v.buffer works only in GRASS7 and only if it uses the GEOS algorithm. A backport to GRASS6 is possible.

[neteler]

(v.buffer2 is v.buffer for a while)

See also #2223

See also #1464

[neteler]

v.buffer has been rebased on GEOS in r59970 which you now need to compile the module.

Please test the updated version of v.buffer in 6.4.svn.

[neteler]

Done in G7, downgrading to G6 where GEOS support was added in r59970.

# G6:
v.buffer input=roadsmajor@PERMANENT output=roadsmajor_buf1_2000 distance=1000

Appears to work, closing (reopen if not).

Some fun performance comparison:

• G6.4.svn: 43.19 sec
• G7.0.svn: 7.59 sec