[GRASS-dev] GRASS GIS 7.0.1 planning: potential backports

[Markus Neteler]

On Mon, Jun 8, 2015 at 3:05 PM, Moritz Lennert
<mlennert at club.worldonline.be> wrote:
> On 08/06/15 14:27, Markus Neteler wrote:
>>
>> On Mon, Jun 8, 2015 at 8:51 AM, Moritz Lennert
>> <mlennert at club.worldonline.be> wrote:
>>>
>>> On 07/06/15 23:10, Markus Neteler wrote:
>>>>
>>>> r62091 - vector/v.select: speed up OP_OVERLAP - Author: mmetz 2014-09-26
>>>
>>>
>>> I've only tested this very briefly, but results seem identical and the
>>> speedup is very significant, so I would plead for backport, but MarkusM
>>> should have the last word on this.
>>
>>
>> ..  I just locally backported, it also needs
>> https://trac.osgeo.org/grass/changeset/62045
>
> Ah, this does look quite heavy, indeed. Has anyone else had time to test ?

Yes, I am testing (and it has been in place since Sept 2014).

Here the comparison, in GRASS 7.0.1svn (nc_spm_08_grass7):~ >

# old version:
v.buffer input=railroads output=railroads_buf20m distance=20

time -p v.select ainput=zipcodes_wake binput=railroads_buf20m
output=zipcodes_wake_railroads operator=overlap
v.select complete. 171 features written to output.
real 125.23
user 82.05
sys 42.61

#########
# new version

v.buffer input=railroads output=railroads_buf20m distance=20

time -p v.select ainput=zipcodes_wake binput=railroads_buf20m
output=zipcodes_wake_railroads operator=overlap
v.select complete. 171 features written to output.
real 6.72
user 4.42
sys 2.03

6.72/125.23 = 0.05366126

It is *quite* faster!

> No news from MarkusM ?

No.

markusN