Re: [gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-10 Thread Jukka Rahkonen
Even Rouault mines-paris.org> writes: > > Jukka, > > I tried to simulate your scenario but couldn't observe any performance > difference (on Linux, but I doubt that the OS would make a difference) when > generating a test DB with 3.7.9 or 3.7.14, and using it with the other > version. > Howe

Re: [gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-09 Thread Even Rouault
Jukka, I tried to simulate your scenario but couldn't observe any performance difference (on Linux, but I doubt that the OS would make a difference) when generating a test DB with 3.7.9 or 3.7.14, and using it with the other version. However it might heavily depend on the number of records that

Re: [gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-09 Thread Jukka Rahkonen
lqt.it> writes: > > On Mon, 9 Jun 2014 07:31:05 + (UTC), Jukka Rahkonen wrote: > > Perhaps this has something to do with different SQLite versions > > included in > > GDAL and Spatialite-gui. My GDAL is v.2.0 Win-64 build from > > gisinternals. > > > > This kind of behaviour is nasty when

Re: [gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-09 Thread a . furieri
On Mon, 9 Jun 2014 07:31:05 + (UTC), Jukka Rahkonen wrote: Perhaps this has something to do with different SQLite versions included in GDAL and Spatialite-gui. My GDAL is v.2.0 Win-64 build from gisinternals. This kind of behaviour is nasty when planning to build well optimized and index

Re: [gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-09 Thread Even Rouault
Hi, This should hopefully be pretty much independant of GDAL itself, but rather tied to the sqlite version used. I can see that gisinternals builds use sqlite 3.7.9. In the news of version 3.7.14 I read "And the query planner has been enhanced to better use covering indices on queries that use

[gdal-dev] GDAL may create slow GPKG/SQLite indexes

2014-06-09 Thread Jukka Rahkonen
Hi, I have a table with 1.2 million rows. Attribute 1 has 320 distinct values (text). Attribute 2 is numeric (area). My test query is this: select * from table where attr_1='009' or area>1000; First index makes query a bit faster ogrinfo test.gpkg -sql "create index test_attr_1_idx on test (att