QGIS » QGIS Application

An update.

I decided to take a look at the connections that were being sent to PostGIS while my script was running using this SQL:

SELECT query FROM pg_stat_activity;

I get three standard lines from Postgres:

"SELECT count(*) FROM pg_attribute WHERE attrelid = 'pg_catalog.pg_proc'::regclass AND attname = 'proargdefaults'"

"WITH configs AS (SELECT rolname, unnest(setconfig) AS config FROM pg_db_role_setting s LEFT JOIN pg_roles r ON r.oid=s.setrole WHERE s.setdatabase=24576)
SELECT rolname, split_part(config, '=', 1) AS variable,        replace(config, split_part(config, '=', (...)"

"SELECT query FROM pg_stat_activity;"

¨
When the script is running with QGIS 2.0 I two lines that seem to be coming from my script. They are:

"CLOSE qgisf0_21"

"SELECT styleQML FROM layer_styles WHERE f_table_catalog='rog_data' AND f_table_schema='' AND f_table_name='(Select int4(row_number() OVER ()) AS id, table.knrgnrbnr, ST_INTERSECTION(table.geom, rog_ar5_sol.geom) As geom FROM (SELECT knrgnrbnr,  (...)"

I never seem to get any more lines than this while the script is running and these disappear each time the loop iterates and I get a new line. By the end of the script I am left with just the first 3 lines.

With Master, I get the Close qgisf0 3 line, followed by the line that starts SELECT styleQML…, but I get 5 of these lines for each loop in my script. And they do not disappear. They just keep adding to the list. When it gets to 100 lines, the script hangs.

Have found this to be a good workaround:

vlayer = QgsVectorLayer(uri.uri(), layerName, "postgres", False)

specifying False for the loadDefaultStyleFlag parameter.

This parameter stops QGIS from checking if there are styles in the database, and thus means that it doesn’t run loads of queries that it then doesn’t close.

QGIS still needs to unload some of the connections that it creates when searching for styles when loading query layers (if you don’t specify the False for the loadDefaultStyleFlag), so I believe the bug is still valid, but this is a workaround for me so that is good. I also wonder if True is the right default for the loadDefaultStyleFlag parameter?

I have no idea if this still is the case. I cant remember which script I was testing it with as I reported this over 4 years ago.

Probably best to close it and someone can report it again if they get the same problem.