Feature request #21279
Processing/OGR: do not use shapefile as format for temp input files
|Pull Request or Patch supplied:||No||Resolution:|
|Easy fix?:||No||Copied to github as #:||29097|
The algorithms from GDAL's toolbox for vector processing still use shapefiles as intermediate format. This results in corrupted fieldnames, since those are limited in the DBF format. Also, field contents might get corrupted if they're longer than 255 chars.
ogr2ogr -f "ESRI Shapefile" C:/Users/tobwen/AppData/Local/Temp/processing_5fe343f547494d5e9fc31fe5cc38c868/680b663b94af46ab9516f2acd4141efd/OUTPUT.shp R:/demo.geojson
Please use GPKG for this.
#2 Updated by Tobias Wendorff over 1 year ago
Sorry, Jürgen, I strongly disagree that this is a feature request only.
The use of shapefile is an intermediate format is creating unexpected results and is destroying the user's results. Furthermore, ogr2ogr actually is one of the most capable applications out there, which can handle GPKG files. Even if I don't like GPKG that much, it has removed limits, we had with prosperity shapefile over the last decades.
I believe patching this is damn easy, you only need to replace
-F Esri Shapefile with
-F GPKG and fix filenames from .shp to .gpkg.
Fixing this bug will save hours of processing, energy and might be even some kittens.
#5 Updated by Tobias Wendorff over 1 year ago
Giovanni Manghi wrote:
is this about specifically the OGR based "convert format" tool?
I do not know whether this question was addressed to me or whether I can answer it at all. As far as I can tell, all the vector tools are affected, f.e. Vector geoprocessing -> Buffer vectors
ogr2ogr C:/Users/tobwen/AppData/Local/Temp/processing_08c1c2be50d4431a82d6d123c3eeec35/ae5cee8d9c484a1986ed5a3db42f1d50/OUTPUT.shp R:/demo.sqlite -dialect sqlite -sql "SELECT ST_Buffer(geometry, 10.0) AS geometry,* FROM 'demo'" -f "ESRI Shapefile"
#8 Updated by Tobias Wendorff over 1 year ago
To prevent oblivion: this is still an issue in QGIS v3.7.0-Master, revision 8c21a9e848
Since latest GDAL (>= 2.5) supports proj6, the use of OGR might be interested to advanced users. Right now, you need to remove all (destroyed) columns, rejoin the original table and write it to a new file.