Lars Hiller Eidnes

I investigated Q21. reorder_joins=1 actually finds a faster join order for it, but for that query, it somehow hurts performance that the join keys are removed from the WHERE clause....

Edit: The issue described below is fixed, so nevermind. ~~There's an annoying issue relating to column names, and I need some help.~~ ~~The current behaviour is as follows:~~ ~~1) When...

I worked around the column name issue by simply storing what the names would be before reordering, and applying those names again later. It meant that TreeRewriter::analyzeSelect(..) and SelectQueryExpressionAnalyzer::appendProjectResult(..) had...

@vdimir @novikd @UnamedRus CI is green for this. Is anything needed from my side to start code review?

> I believe I have this build installed, how does it work exactly? The optimization doesn't trigger because you're using the `JOIN .. USING (..)` syntax. See the comment in...

@alanpaulkwan ```sql SELECT count() FROM crsp_202301.dsenames AS dsf INNER JOIN crsp_202301.dsf AS dse ON dsf.permno = dse.permno WHERE (date >= start) AND (date = start) AND (date

> Running it twice it disappears. Cool! Good to hear :) > I'll keep experimenting with join patterns and report back with anything weird I see (is there a better...

Happy to hear that @alexey-milovidov I'm not sure exactly what changes you have in mind for `calculateCost`. Could you elaborate? The code I used to find these coefficients is in...

> which stream left table and put whole right table in memory. Ok, that makes sense then, thanks for explaining.

> @larspars Hey man, will you continue this work? Since it is a great feature! I'd be happy to give it another go. But I would need some guidance on...