rzo1 commented on PR #305:
URL: https://github.com/apache/openjpa/pull/305#issuecomment-6851420296

   DB specific failures, pushed in 627f9108a, 7e242c709 and a277c3edc:
   
   - `TestRelationFieldAsPrimaryKeyAndForeignKey` (MySQL, MariaDB): as you 
suggested, the String ids of the six entities get `@Column(length = 64)`. 
`D4_REL`'s PK was 4 × `VARCHAR(255)` = 4080 bytes in utf8mb4, over InnoDB's 
3072. On master the test silently disabled itself on that error, so it actually 
runs there now.
   - `TestMappingToolRefresh` (MySQL, MariaDB, PostgreSQL): not INT vs BOOL :) 
it is order dependent. `TestJDBCSchemaManager` leaves `EntityBoolChar` with the 
native boolean column (`BIT(1)` / `BOOLEAN`), and `SchemaAction='add'` never 
changes an existing column, so inserting `'1'` as `STRING_10` fails. Stage 1 
now uses `'drop,add'`.
   - `TestRemoteCommitTypeNames` (the one you could not reproduce): the query 
cache keeps its listeners in a *weak* set, and the test registered a lambda 
without keeping a reference, so a GC between the two events dropped it 
(reproducible with a `System.gc()` in between). The listener is now held in a 
local and removed in `finally`; same in `TestRemoteCommitUnloadedSubclass`. 
Latent on master as well.
   
   All verified on MySQL 8.4, MariaDB 12.3, PostgreSQL 18 and Derby.
   
   Random on master and alphabetical (or a fixed seed) on PRs sounds good to me.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to