You probably refer to one of the demos on our landing page?
I like how you describe the problem. You're absolutely right that SQL seems easy but it's these edge cases that make it hard to get right. Joining metrics with a date spine is definitely a good practice to avoid missing date periods.
I think we could/should teach Dot to do that in the future. It should at least be a feature you can turn on as the data team.
Indeed, not sure how I ended up there but did on mobile, commented here.
> You're absolutely right that SQL seems easy but it's these edge cases that make it hard to get right
SQL/data analysis is endlessly pesky! I assume it would be easier to spot on tighter increments like "minutely" or "hourly"
> It should at least be a feature you can turn on as the data team.
Some might want the missing points, others wont - sounds like a good option (but id default to "enabled", each to their own though)