tokoko commented on issue #13:
URL: https://github.com/apache/arrow-adbc/issues/13#issuecomment-4687600393

   > I'm not 100% sold on the value of a driver which just reads Parquet files, 
but I know better than to think I can anticipate other people's interesting (or 
unusual :P) use cases. I^H Claude might prototype one this weekend.
   
   I was just talking to David about this... 😆 I dislike the "just use 
duckdb/DF" approach even though that's technically very possible of course. If 
you take that argument further, you can also argue that postgres driver is not 
necessary as well since duckdb postgres extension exists. there's really no 
reason for adbc to exist if we're fine with treating duckdb as an api layer + 
local processing engine. I think we are disregarding the fact that the point of 
adbc is that it's just an api spec and that matters. A particular reader is 
free to bundle whatever it needs inside the driver, but shouldn't rely on 
something on top if that makes sense.
   
   From a practical perspective, you can't really distribute a parquet or 
iceberg read operation (at least easily) if what you're relying on is a local 
compute engine, while for native adbc drivers it's a straightforward task. same 
goes for distributed write which isn't really possible with adbc either today, 
but at least we can fix that.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to