... aka "the SQLite Surge".
- Invalid data 6 passes, 1 fail, and 0 exceptions
- Drupal error handlers 27 passes, 1 fail, and 0 exceptions, see #543594: Exceptions in PDO doesn't bubble up
| Comment | File | Size | Author |
|---|---|---|---|
| #1 | sqlite-test-report.txt | 1.74 MB | damien tournoud |
Comments
Comment #1
damien tournoud commentedHere is the full test log.
Comment #2
damien tournoud commentedRemoved the admin menu failures... that are obviously not core.
Comment #3
Paul Natsuo Kishimoto commentedIs this fixed if #543594: Exceptions in PDO doesn't bubble up is fixed?
Comment #4
sun.core commentedIndividual issues can be critical, but let's move meta issues out of the critical queue, please.
Comment #5
damien tournoud commentedHere are the remaining failures as of today:
This is probably a borked test.
This is because SQLite doesn't trim fixed-precision data by default. I'm unsure how to fix that (except by removing the test).
This is due to #715108: Make Merge queries more consistent and robust.
This is because of a difference on how MySQL and SQLite handles the notion of "affected row" when there is a float column. I'm pondering.
Comment #6
damien tournoud commentedWith #716006: Schema tests fail on SQLite, #714964: ActionLoopTestCase is fragile and #715108: Make Merge queries more consistent and robust applied, we are down to 3 failures!
Comment #7
andypostSchema tests fixed #720620: indexExists() for pgsql and sqlite does not prefix tablename
Comment #8
andypost#754192: Transaction support for old sqlite versions
#357470: SQLite should use ON CONFLICT ROLLBACK insert and update queries
Comment #9
mfer commentedThere are currently a number of failing SQLite tests. Here are the current failures:
That's 61 failures and 16 exceptions.
I marked this as critical until the issues from this are found and created. D7 is ready to go when all the critical issues are gone. The SQLite test failures are critical.
Comment #10
damien tournoud commentedThis is the meta-issue, it shouldn't be marked critical.
If you want to help, please open *separate* issues and link them back here.
This is just due to a granularity issue. I have a patch.
I never saw those. I'm tempted to say that there is something wrong in your environment.
Comment #11
mfer commentedSome of these are not related to SQLite. But, they are happening on several systems. On systems older than PHP 5.2.5 some of the problems go away. Digging deeper into this.
Comment #12
mfer commentedThe core filter tests are unrelated and an entirely different problem. See #755758: check_plain throwing warnings for more detail.
Comment #13
damien tournoud commentedAs of today:
Comment #14
damien tournoud commentedThat was because my version of SQLite didn't support SAVEPOINTs. Upgrading to a newer version fixed that.
This is because the base implementation of InsertQuery doesn't actually rollback the transaction if one insert fails.
Comment #15
damien tournoud commentedWith the following patches applied:
#851166: SQLite schema implementation doesn't support qualified tables
#851168: db_find_tables() expects tables to be prefixed, inconsistent implementation and documentation
#851110: InsertQuery::execute() needs to rollback the transaction if one insert fail
#851140: DatabaseSchema::getPrefixInfo() is *severely* broken
#850852: Fix transaction failure and allow concurrent testing on SQLite
Here are the three remaining failures tonight:
Comment #16
damien tournoud commentedWith the following patch stack:
#851698: DatabaseSchema_sqlite severely broken
#851168: db_find_tables() expects tables to be prefixed, inconsistent implementation and documentation
#851110: InsertQuery::execute() needs to rollback the transaction if one insert fail
#851140: DatabaseSchema::getPrefixInfo() is *severely* broken
#850852: Fix transaction failure and allow concurrent testing on SQLite
#809600: Stop using bit-wise operators for {file_managed}.status
#851590: SQLite should "stringify" fetches
#851602: FieldCrudTestCase::testUpdateField() does not test changing the schema
We get 100% pass! Yay!
Comment #17
steven jones commentedCan this issue be closed now? It seems very much out of date?
Comment #18
bfroehle commentedDrupal 7 Continuous Integration shows one remaining fail which is addressed in #1001242: DBTNG support for EXISTS conditions.
Comment #19
bfroehle commentedAll tests pass for me... :)