On 23.01.2019 04:34, Branko Čibej wrote:
> On 22.01.2019 23:10, Branko Čibej wrote:
>> On 22.01.2019 22:52, Branko Čibej wrote:
>>> [Moving to the correct list.]
>>>
>>> On 22.01.2019 22:18, Thomas Deutschmann wrote:
>>>> Hi,
>>>>
>>>> in Gentoo Linux, wc-queries-test test is failing (verified
>>>> against v1.9.7-1.11.1:
>>>>
> [...]
>>>>> svn_tests: E200006: STMT_DROP_TARGETS_LIST: Uses sqlite_master with only 0 index component: ((null))
>>>>> DROP TABLE targets_list
>>>>> svn_tests: E200042: Additional errors:
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> svn_tests: E200035: |SCAN TABLE sqlite_master
>>>>> FAIL: wc-queries-test 3: test query expectations
>>>>> PASS: wc-queries-test 4: test query duplicates
>>>>> PASS: wc-queries-test 5: test schema statistics
>>>>> PASS: wc-queries-test 6: verify queries are parsable
>>>>> END: wc-queries-test
>>>>> ELAPSED: wc-queries-test 0:00:00.167083
>>>> After some research we found the problem causing the test failure:
>>>>
>>>> In Gentoo we are building sqlite with
>>>>
>>>>> -DSQLITE_ENABLE_STMT_SCANSTATUS
>>>> option set. If we disable that option, the test is passing.
>>>>
>>>> Downstream-bug: https://bugs.gentoo.org/605438
>> I've filed this as issue #4804, it looks like a bug in our tests.
>
> I was able to reproduce this with the latest (amalgamated) SQLite on
> macOS — so, not using get-deps.sh, as the version that downloads is too
> old. I've written up my analysis in issue #4804; the short story is that
> adding the compile-time option mentioned above adds output to 'EXPLAIN
> QUERY PLAN' that our test doesn't like.
>
> Bert, you wrote most of this query test code; would you mind taking a
> look at this issue?
I decided to "fix" this by ignoring it ... see r1852013. I think the fix
is correct, given what the test is supposed to do.
-- Brane
Received on 2019-01-24 11:59:41 CET