4 ms·

Times up. LLMs are uniquely suited to convert data requests to SQL at compile-time. All ORMs will converge on this feature “scan your table layout and generate

Times up. LLMs are uniquely suited to convert data requests to SQL at compile-time. All ORMs will converge on this feature “scan your table layout and generate smarter queries with Knexjs Pro+” - one can see it coming. My mother raised me as a broke single mother because she could write SQL. I bought a home in my California home town because I could write good SQL and tune my.cnf. I suppose my toddler will never SELECT * FROM. I wish it weren’t so, but it’s always felt a bit silly, hasn’t it? “Hey Siri, People own many Widgets and belong to one Team” is coming. Why would we want the future of computing not to obscure database operations? Why should our grandchildren care about shards? Let it all die.

willvarfar 2y ago
The reason the big focus on LLMs generating SQL was that SQL is somewhat more natural language like? It doesn't mean that LLMs generate sensible or even correct SQL. I work at a big company with big data and sometimes trawl the query logs to speed things up and recently noticed an uptick in WTF queries that, I suspect, have had some AI 'assistance' ;)
otabdeveloper4 2y ago
Absolutely not. For example, there is literally nothing at all in common between a Postgres query and a Clickhouse query except some very basic surface-level syntax. (And even then lots of things are incompatible, e.g., count(*) vs count())
- zX41ZdbW 2y ago
  You don't have to write `count()` in ClickHouse - `count(*)` is also ok. ... while `count()` makes more sense from the engineering perspective: it is an aggregate function that takes zero number of arguments. I personally prefer `count()`, but when my friends write `count(*)` or even `COUNT(*)`, they are still my friends.
  - hodgesrm 2y ago
    The count() invocation is a small but excellent feature in ClickHouse. There are many other great dialect extensions. -If combinators are another great example. [0] SQL would be a better language if there had been more innovation focused on code usability over its long history. [0] https://clickhouse.com/docs/en/sql-reference/aggregate-functions/combinators#-if https://clickhouse.com/docs/en/sql-reference/aggregate-funct...

willvarfar 2y ago

The reason the big focus on LLMs generating SQL was that SQL is somewhat more natural language like? It doesn't mean that LLMs generate sensible or even correct SQL. I work at a big company with big data and sometimes trawl the query logs to speed things up and recently noticed an uptick in WTF queries that, I suspect, have had some AI 'assistance' ;)

otabdeveloper4 2y ago

Absolutely not. For example, there is literally nothing at all in common between a Postgres query and a Clickhouse query except some very basic surface-level syntax. (And even then lots of things are incompatible, e.g., count(*) vs count())

zX41ZdbW 2y ago
You don't have to write `count()` in ClickHouse - `count(*)` is also ok. ... while `count()` makes more sense from the engineering perspective: it is an aggregate function that takes zero number of arguments. I personally prefer `count()`, but when my friends write `count(*)` or even `COUNT(*)`, they are still my friends.
- hodgesrm 2y ago
  The count() invocation is a small but excellent feature in ClickHouse. There are many other great dialect extensions. -If combinators are another great example. [0] SQL would be a better language if there had been more innovation focused on code usability over its long history. [0] https://clickhouse.com/docs/en/sql-reference/aggregate-functions/combinators#-if https://clickhouse.com/docs/en/sql-reference/aggregate-funct...