Sven Klemm

Can you reproduce the result when the index definitions are exactly the same?

When adjusting the index definitions, the index size is the same after the fill script has been run. Here is the diff i used for the scripts: ``` diff --git...

Have you considered not storing these as dimensions. This seems to be quite an invasive change and it means that all existing dimension code needs to be checked whether it...

Hmm there is definitely something off. I boiled down the example a bit further: ``` sven@dev[82029]=# WITH data (time) AS (VALUES ('2024-01-31 02:00:00+00'::timestamptz), ('2024-01-31 02:00:00+00'::timestamptz) ) SELECT time_bucket_gapfill('1h'::interval, time, 'Europe/Stockholm')...

generate_series does produce the expected result: ``` sven@dev[82029]=# SELECT generate_series('2024-03-31T0:00Z','2024-03-31T02:00Z','1h'::interval) at time zone 'Europe/Stockholm'; timezone --------------------- 2024-03-31 01:00:00 2024-03-31 03:00:00 2024-03-31 04:00:00 (3 rows) ```

> Unfortunately it will not work because we use low level functions to update catalog tables and we only check for nulls, so no other constraints will be enforced when...

Issue seems to be resolved. Feel free to reopen if you still have problems.

Looking at the test output i dont think your change actually fixes the problem

On self-hosted you can manually delete the offending entries as a workaround. ``` DELETE FROM chunk_constraint WHERE chunk_id = 1; ```

> What exactly is broken? I reproduced it but I have no idea what's going on... Theoretically, the `make_restrictinfo` should take care of creating the proper `orclause` as well. Not...