I'm trying to use the Python API to create a temporary map, add a single point feature, then "print" (export) to file. My problem is that when I print the map, the feature doesn't appear. However, if I save the map and open it in a browser, the feature is visible.
I think I've tracked down the behavior causing this: adding features from a query results in a feature collection on the output map. The print() call renders features from the collection in order until it encounters one with a value in a GUID field. It fails to render that feature and all subsequent features in the collection. However, the features ARE there on the webmap. If you view the map in a browser or examine the JSON, they’re present.
- Removing all values from all GUID fields results in all features rendering in the print correctly.
- Adding parameter out_fields=’OBJECTID’ to the layer query results in features rendering in the print correctly.
- Adding a GUID value to a text field does not affect rendering.
- The name of the GUID field doesn’t appear to matter. My original data had two GUID fields (ParentGlobalID and parent_feature_globalid). A value in either field resulted in failure to render.
- The number of GUID fields doesn’t appear to matter. I’ve tested with hosted feature service layers having two, one, and no GUID fields. Any value in any GUID field resulted in failure to render.
- The number of features doesn’t appear to matter. As stated above, it fails to render the first feature with a GUID value and all subsequent features. I began testing with ~50 features and reliably produced this behavior.
- It only happens when you add features from a query, which results in a featureCollection in the output webmap. If you add the same data that failed to render as a featureCollection to the map as a featureLayer instead, it will render correctly.
- My original data had relationships with other layers in the same service. I tested other feature service layers with and without relationships. Having relationships that reference the GUID field doesn't appear to be a condition of the bug.
For my purposes at least, this bug is easy to work around. I don't need the GUIDs for my print, so I just added parameter out_fields='OBJECTID' to my query so that I don't offend the print() call's delicate sensibilities with those big, scary GUIDs.
Sponsor me ESRI I practically work for you at this point.
