A single materialized view is generated for each exposure_model. This view has two geometry columns - geom for a coarse view; full_geom for a detailed view. If the underlying data does not have detailed geometry the view populates full_geom with the same geometry as the one in geom.
Removal of unneeded fields from the materialized view, namely those that came from the cost, model_cost_type and occupancy tables
Addition of two flags to the management command:
-a - do not VACUUM the database. This is mainly useful for development, it saves some time during runs
-n - print the SQL commands for creating each materialized view and also the plpgsql function to stdout. Do not run these commands, only show them
A slight refactor of the plpgsql function in order to cut back on some local variables
The overall performance of the view generation and population has been improved. The ingestion of the test data is now taking ~16 minutes on my local machine (while previously it took ~40min)
This PR is connected to #15
It features the following modifications:
A single materialized view is generated for each exposure_model. This view has two geometry columns -
geom
for a coarse view;full_geom
for a detailed view. If the underlying data does not have detailed geometry the view populatesfull_geom
with the same geometry as the one ingeom
.Removal of unneeded fields from the materialized view, namely those that came from the
cost
,model_cost_type
andoccupancy
tablesAddition of two flags to the management command:
-a
- do not VACUUM the database. This is mainly useful for development, it saves some time during runs-n
- print the SQL commands for creating each materialized view and also the plpgsql function to stdout. Do not run these commands, only show themA slight refactor of the plpgsql function in order to cut back on some local variables
The overall performance of the view generation and population has been improved. The ingestion of the test data is now taking ~16 minutes on my local machine (while previously it took ~40min)