I store multiple geometry columns in a single table. I'd like to be able to use the PostGIS Reader to decide which geometry I'd like to pull out and transform.
PostGIS Reader should handle multiple geometry columns in a table
FME Form
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.
+1 Handling multiple geometries is a need.
I've encountered a bug when dealing with Postgis mutliple geometries table which requires to specify which geometry is taken into account, otherwise FME loses geometry :
https://github.com/qwat/qwat-data-model/issues/265
This may help:
https://knowledge.safe.com/articles/24702/handling-data-that-contains-multiple-geometry-colu.html
However yes, a more intuitive way would be greatly appreciated by many of us.
See also duplicate Idea: https://knowledge.safe.com/idea/117563/native-support-for-postgresql-multiple-geometry-co.html
One of the advantages of PostGIS is that it can handle multiple geometry columns in the same table. For example and a polygon area of interest, then the gradual erosion of that AOI which regard to image collection, etc. Should not have to choose one or the other, in my opinion.
I now too have this challenge whereby the table holds multiple geometries. Coming from Oracle Spatial and seeing how easy it works with FME I was surprised Safe has not implemented the same for PostGIS
I realise this idea relates to the PostGIS Reader, but it would also be good if PostGIS Writer could write to multiple geometry columns.
I tried the workaround at https://community.safe.com/s/article/template-write-multiple-geometry-types-to-an-exist, but writing the second geometry column is extremely slow, as bulk insert for this writer is switched off during translation (the following warning message is received):
POSTGIS Writer: Bulk Insert is set to Yes, but a conflicting Feature Operation has been detected. Feature type 'rc_srb.srb_01' has a Feature Operation of 'Update'. Bulk Insert is now being set to No, which can cause insert operations to be much slower
As a result, the timings for writing a dataset of 130,000 features is as follows:
All the votes have been transferred into this idea.