Opened 5 years ago
Closed 5 years ago
#4523 closed defect (fixed)
Upgrade 2.1.9 to 3.0.0rc1 - can not drop function st_union(geometry) because other objects depend on it
Reported by: | robe | Owned by: | strk |
---|---|---|---|
Priority: | medium | Milestone: | PostGIS 3.0.0 |
Component: | build | Version: | master |
Keywords: | Cc: |
Description (last modified by )
I'm not sure this test is fair -- dronie is showing this error going from 2.1.9 to 3.0.0rc1
https://dronie.osgeo.org/postgis/postgis/669
Died at ../run_test.pl line 1588. failed (Error encountered altering EXTENSION POSTGIS: /tmp/pgis_reg/regress_log) ----------------------------------------------------------------------------- CREATE EXTENSION WARNING: unpackaging raster WARNING: PostGIS Raster functionality has been unpackaged HINT: type `SELECT postgis_extensions_upgrade(); to finish the upgrade. After upgrading, if you want to drop raster, run: DROP EXTENSION postgis_raster; ERROR: cannot drop function st_union(geometry) because other objects depend on it DETAIL: view upgrade_view_test depends on function st_union(geometry) HINT: Use DROP ... CASCADE to drop the dependent objects too. CONTEXT: SQL statement "DROP AGGREGATE IF EXISTS ST_Union (geometry)" PL/pgSQL function inline_code_block line 15 at EXECUTE ----------------------------------------------------------------------------- make[1]: *** [check] Error 1 make[1]: Leaving directory `/drone/src/regress/core' make: *** [check-regress] Error 2 make: Leaving directory `/drone/src/regress' FAIL: upgrade 2.1.9--3.0.0rc1
Can we not test 2.1.9 as that's long deprecated already and besides we know that ST_Union changed. It's changed I think 3 times since 2.1.9 so this is an expected failure
Change History (7)
comment:1 by , 5 years ago
Description: | modified (diff) |
---|
comment:2 by , 5 years ago
comment:4 by , 5 years ago
On the other hand, this block in postgis_upgrade_for_extension.sql.in
seems to drop the aggregate in all cases for Pg < 12.
$postgis_proc_upgrade$ BEGIN IF current_setting('server_version_num')::integer >= 120000 THEN EXECUTE $postgis_proc_upgrade_parsed_def$ CREATE OR REPLACE AGGREGATE ST_Union (geometry) ( sfunc = pgis_geometry_accum_transfn, stype = internal, parallel = safe, finalfunc = pgis_geometry_union_finalfn ); $postgis_proc_upgrade_parsed_def$; ELSIF 205 > version_from_num OR ( 205 = version_from_num AND version_from_isdev ) FROM _postgis_upgrade_info THEN EXECUTE 'DROP AGGREGATE IF EXISTS ST_Union (geometry)'; EXECUTE $postgis_proc_upgrade_parsed_def$ CREATE AGGREGATE ST_Union (geometry) ( sfunc = pgis_geometry_accum_transfn, stype = internal, parallel = safe, finalfunc = pgis_geometry_union_finalfn ); $postgis_proc_upgrade_parsed_def$; END IF; END $postgis_proc_upgrade$;
Is your test on the durability of ST_Union a little too optimistic perhaps strk?
comment:5 by , 5 years ago
Isn't this hitting the change done in 2.5 (internal type) and not the one just reverted?
OTOH, I'd be ok with dropping this test. Things work as they should with PG12.
comment:6 by , 5 years ago
I've dropped this test in r17882, close this out unless someone can prove to me that we are testing behaviour we ever actually had.
comment:7 by , 5 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
Yah I think its a given if you upgrade prior to 12 and you used aggs in your functions, your best bet is to move to 12.
It doesn't look like it changed earlier:
So.. maybe Paul just forgot to remove the
DROP AGGREGATE
line from upgrade_before ?