Hi, I am currently looking at Cairo with the intent of using it as a map "rendering engine". I work at a company making GIS applications, which typically has a map on the screen. Our current rendering is done by using win32 GDI, but now we are looking for a better way. I am looking at several alternatives, but I find Cairo quite interesting in many ways. Before I start hammering with coding & trying things out, I'd like to hear your opinions. In our apps, we typically have 5-20 (in some cases less more) layers. Each layer then consists of speciffic geographical objects, for example - height curves (simply polylines) - buildings (polygons, majority 4-20 vertexes, sometimes more) - roads (sometimes just polylines, but they may also be polygons with 1000's of vertices) - rivers/sea etc (often huge polygons, with "islands"/holes and several thousand vertexes) - texts, or points with truetype/pixmap symbols - background photos, ortho photos taken from airplanes In addition, - polygons are typically parametrically hatched, or filled with some kind of pattern. - lines may come in many variants. Some have just dash/dot patterns, orhers may have additional figures along it (like direction arrows, circles etc) The app typically starts with an initial zoom. The user may suddenly move to a different locations, which triggers reading new geometries from the databases and then renders it. This means that the app may not keep a static/permanent model in RAM all the time. One render may contain, in "heavy" cases, 10.000 polygons of various types, and 30.000 lines.. or even more. Does this sound like a task for Cairo ? And, how is Cairo with memory usage ? I am eager to try this out, but still, any comments/suggestions are welcome ! Oyvind