Zach Musgrave

Jason's query above now dies with a different error: ``` > select from_a, from_x, from_y, to_a, to_x, to_y, from_commit, to_commit, diff_type from dolt_diff_a order by to_x; column "from_a" could not...

Max, please close this out when your work with removing filters on index lookups is done

There should always be a single author for a particular commit, even if multiple people committed to a branch that got merged. This should be doable.

Same customer reports consistent slowness on a read replica backed by GCS. Database in question has 1200 branches. Configuration: ``` | Variable_name | Value | +--------------------------------------+--------+ | dolt_allow_commit_conflicts | 0...

This PR should substantially address these latency problems, now consistently

Problem is limited to new tables in the diff. Existing tables with diffs filter fine.

Issue seems to be that the natural join puts the query plan into a shape that it can't be optimized by the join optimizer. Same query rewritten to not use...

This is a relatively large feature and will take quite a bit of work to implement. I'm adding it to the roadmap.

We would like to support this and it's on our roadmap, but we don't have an ETA for when it will become available. In the meantime, you can use this...

It's harder. It's not just syntax, it's got query planner logic too.