Le vendredi 16 octobre 2015 20:14:44, Hermann Peifer a écrit : > On 2015-10-16 18:20, Even Rouault wrote: > > Le vendredi 16 octobre 2015 17:57:10, Hermann Peifer a écrit : > >> On 2015-10-12 11:28, Even Rouault wrote: > >>> I've changed to '%.15g' formatting for WKT in > >>> https://trac.osgeo.org/gdal/ticket/6145 > >> > >> Hi Even, > >> > >> Why did you choose '%.15g', rather than '%.17g'? The latter format > >> should guarantee a lossless binary->decimal->binary round-trip, whereas > >> the former one might also do -- or not. I don't think that any precision > >> loss between '%.15g', and '%.17g' would result into "a real life > >> problem", I am asking just out of curiosity. > > > > Hi Hermann, > > > > re'adding the list in CC. > > > > Your point is valid. My idea was to avoid making WKT returned by the > > modified version potentially longer than it is currently > > I am pretty sure you know: '%.17g' could also result into shorter string > values, compared to the earlier '%.15f', as in the below (admittedly > unfair) example.
Yes, that's intended that the output is potentially now shorter than before. Previous formatting outputed too many figures, especially for projected coordinates. > > Hermann > > $ printf "%.15f\n" 50.375 > 50.375000000000000 > $ > $ printf "%.17g\n" 50.375 > 50.375 -- Spatialys - Geospatial professional services http://www.spatialys.com _______________________________________________ gdal-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/gdal-dev
