It'll be good to have sql query-able metadata information_schema views/objects for automation etc
As an sql dev, i write and execute code many hundreds of times a day
one feedback loop i've become used to on other platforms to is to be able to write an sql select query, hit execute (and irrespective of how many rows are in the final dataset), and start seeing the rows streaming within a few sec - basically as soon as the query begins executing. Does that make sense?
an example
0sec: submit sql query
1-10 sec; generate plan, send to workers
11-100sec; query is executed and data sent to ide
== 15-105 sec: some rows start appearing in ide, continue to stream in until query is finished
so I start seeing the result of my query after ~15 sec, and long before it's finished returning all data
But in workbench j, other jdbc sql clients what I see is:
0sec: submit sql query
1-10 sec; generate plan, send to workers
11-100sec; query is executed and data sent to ide
101-130 sec: IDE churns thru resultset to serialise it into a visual/grid all in one go
131 sec: i see all the data
difference: see the first rows of data after ~15sec vs 2+minutes
multiply that a couple hundred times a day and it adds up
I find this useful for
1: implicitly gauging the performance of my queries
3. agile data dev - i can write a query w/o spending time thinking about hand crafting exclusions/filters or how many rows i might need
2: running tests/debugging prod code output interactively - because the data returns as soon as some data is available and i can kill the query once i know enough