Seth Wiesman

Results 19 comments of Seth Wiesman

The more I test with larger postgres instances, the more this feels like a hard requirement. For example, tpch scale 10 is 10GB but the regions table only has 5...

Also linking this issue @frankmcsherry indicated may be related. https://github.com/MaterializeInc/materialize/issues/6591

> Do we have more clarity now on what the SQL interface would look like for these kinds of multiplexed sources? CREATE SOURCES is probably the wrong direction. There should...

@morsapaes that's a great point. What about something like: ``` CREATE SOURCE mz_source FROM POSTGRES CONNECTION pg_connection PUBLICATION 'mz_source' FOR ALL TABLES CAST foo.my_enum AS string; ```

Just a quick ping that if we want to do this, it should be picked up soon :)

I just realized another reason to make CAST explicit. When we add other direct db sources, such as MySQL, there may be types without 1-1 mappings or types we never...

@uce we can make that work. During EA, I will control onboarding and can ensure we don’t bring any new pg source users onboard until this issue is resolved

@morsapaes I believe I've addressed all the review comments. I have run this locally under a few different scenarios and feel pretty good about the correctness. I do need some...

@morsapaes happy to be opinionated. What are you thinking? Force sources / sink to be in a different schema than indexes / mat views? Different clusters even?