Frank Chen

Hi @clintropolis @suneet-s , Could you review this PR at any time you're convenient ? Since this PR is a little large, I think the SQL problem could be separated...

> Sorry, I will try to get to this soon! I think I have a similar problem to solve with complex types in a different thing I'm working on, so...

@suneet-s Thanks for your review. I will address all the comments you left and update this PR later this day.

> @FrankChen021 To help with reviewing this PR, could you update the PR description to include some notes how you chose to implement the solution for this. For example, it...

> > SQL query on re-indexed columns with double/float/long first and last aggregators WON'T work. This involves some changes in complex type handling which might be better in another PR....

> @FrankChen021 - have you run any performance tests/benchmark for ingestion time rollup? That will be handy to rule out any perf bug. I have not. But I will. Thanks...

Hi @abhishekagarwal87 , the changes in this PR was previously used in our team. I'm not working on that project now, so I don't have large chunk of time to...

How about using `APPROX_QUANTILE_DS` to calculate quantile?

You're right, this is due to "()" in the name of a json node. And Because this is caused by [JsonPath](https://github.com/json-path/JsonPath) which is used Druid to flatten json object, I...

The `maxOnDiskStorage` in your json is a type of string, please remove the double quotation marks around the number 100. Of course from the code side, I think it can...