Aleksandr Efimov has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/24534 )

Change subject: IMPALA-15161: Add GEOMETRY type + make WKB_EXPERIMENTAL default
......................................................................


Patch Set 17:

(3 comments)

Looked at this from the geospatial test side - the FE parts are well covered by 
the earlier rounds. Three notes on the test reshuffle.

http://gerrit.cloudera.org:8080/#/c/24534/17/tests/custom_cluster/test_geospatial_library.py
File tests/custom_cluster/test_geospatial_library.py:

http://gerrit.cloudera.org:8080/#/c/24534/17/tests/custom_cluster/test_geospatial_library.py@91
PS17, Line 91:   def test_wkb_experimental_extra(self, vector):
This one and test_esri_geospatial_functions_extra above run the same file, 
QueryTest/geospatial-esri-extra, and the class-level with_args starts the whole 
class with --geospatial_library=HIVE_ESRI, so it runs twice in the same mode.

The other half is that geospatial-esri-extra is gone from 
test_geospatial_functions.py, so under the new default mode it is not covered 
at all - before this patch it ran in both modes (ESRI through query_test, WKB 
through the custom-cluster method whose name survived here). Dropping this 
method and adding the file back to the WKB suite would restore the pair.


http://gerrit.cloudera.org:8080/#/c/24534/17/tests/custom_cluster/test_geospatial_library.py@94
PS17, Line 94:   def test_wkb_experimental_serialization(self, vector):
Running the serialization and relations files in both modes makes sense to me - 
ST_AsBinary has a different implementation behind it in each. But the names and 
the comment still describe the old placement: this sits in a class that starts 
with HIVE_ESRI, while the comment says "WKB is the inter-function serialization 
format in this mode". Renaming these two to esri_* would keep the next reader 
from reading the class header and the method name as contradicting each other.

Separately, test_relations_table lost its @pytest.mark.execute_serially on the 
way over (line 99), and so did test_hive_esri (line 55, formerly test_enabled). 
geospatial-relations-table loads a table, so if that mark was doing real work, 
this is a quiet change.


http://gerrit.cloudera.org:8080/#/c/24534/17/tests/query_test/test_geospatial_functions.py
File tests/query_test/test_geospatial_functions.py:

http://gerrit.cloudera.org:8080/#/c/24534/17/tests/query_test/test_geospatial_functions.py@33
PS17, Line 33:         v.get_value('table_format').file_format == 'text')
The comment right above still says "Tests do not use tables at the moment, skip 
other fileformats than Parquet", while the constraint now selects text - and 
the suite does use tables now, through geospatial-relations-table and through 
geometry-type with unique_database. Worth saying in the comment what text is 
chosen for.



--
To view, visit http://gerrit.cloudera.org:8080/24534
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I0ff29cb15ab45f3899bf453041e605a32f195c27
Gerrit-Change-Number: 24534
Gerrit-PatchSet: 17
Gerrit-Owner: Csaba Ringhofer <[email protected]>
Gerrit-Reviewer: Aleksandr Efimov <[email protected]>
Gerrit-Reviewer: Arnab Karmakar <[email protected]>
Gerrit-Reviewer: Balazs Hevele <[email protected]>
Gerrit-Reviewer: Csaba Ringhofer <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Jason Fehr <[email protected]>
Gerrit-Reviewer: Peter Rozsa <[email protected]>
Gerrit-Comment-Date: Tue, 25 Aug 2026 17:04:23 +0000
Gerrit-HasComments: Yes

Reply via email to