On Thu, 17 Mar 2011 17:08:43 -0500, Cline, Royce L. wrote > Did further testing. The problem is a power function of the number > of layers. > > QGIS 1.7 r15529M. > # layers load time quit time > 40 2.3s 1.6s > 60 3.6s 3.4 s > 68 4.8s 5.3s > 76 6.2s 8.9s > 84 7.9s 19.6s > 92 10.9 36.7s > 100 14.1s 57.7s > 108 18.8s 82.2s > > QGIS 1.6 > # layers load time quit time > 108 3.6s instantaneous >
Hi List, while debugging the above issue noticed by Royce, I casually discovered a major issue affecting any recent (1.7.0) QGIS version. Please note well: such issue affects any operating system, and isn't at all specifically related to SpatiaLite: it's a much more general flaw affecting Map Canvas and Map Layer Registry handling. I've prepared a very small test suite you can use to reproduce this issue in the easiest way: http://www.gaia-gis.it/merano_shp.zip ---------------------- shortly said: the test suite contains a SpatiaLite DB storing some 110+ layers; anyway each single layer simply contains one or two features only, so it is an ultra-light weighted dataset. and I've included Shapefiles as well, exactly corresponding to DB layers. the same identical slowness affects both SpatiaLite and Shapefiles (and I strongly suspect, any other vector datasource): it looks like an exponentially increasing time is required to add (or remove) any further layer. so adding very few layers requires a quite unnoticeable time: but adding several tenths (or hundredths) layers at the same time requires an intolerable time. please note: the number of features actually stored into each single layer has very little effect. I'll inform you about any further advancement. bye Sandro _______________________________________________ Qgis-developer mailing list Qgis-developer@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/qgis-developer
