sq should be able to work with datasources defined in-line

[neilotoole]

Currently one must do sq add DS and then reference the data source in a later command. It should be possible to refer to data sources in-line, e.g.

sq '@myhandle:mysql://user:pass@localhost:3306/mydb1 @myhandle.users | .uid, .email'

# or (without specifying a handle)

sq '@:mysql://user:pass@localhost:3306/mydb1 .users | .uid, .email'

[neilotoole]

This is still a possibility. The post recent proposed syntax on this would be:

$ sq '@(mysql://user:pass@localhost/sakila)' | .actor | .[0:10]

$ sq '@(./path/to/sakila.xlsx).actor'

However, for file-based formats there is less of a need for this anymore due to the ability to do:

$ cat sakila.xlsx | sq .actor

[c-nv-s]

I want to also mention that as a Postgres user, another reason why this feature would make life a lot easier and more versatile is by facilitating the querying of different schemas on the same database more easily.

at present you also have to add each schema as a different src i.e.

sq add 'postgres://sakila:p_ssW0rd@localhost/sakila?search_path=customer1'
sq add 'postgres://sakila:p_ssW0rd@localhost/sakila?search_path=customer2'
sq add 'postgres://sakila:p_ssW0rd@localhost/sakila?search_path=customer3'

whereas it would be a lot easier sometimes to just call

sq '@(postgres://sakila:p_ssW0rd@localhost/sakila?search_path=customer1)' | .customer_name
sq '@(postgres://sakila:p_ssW0rd@localhost/sakila?search_path=customer2)' | .customer_name

...or maybe overriding the schema with a flag could also work

sq add 'postgres://sakila:p_ssW0rd@localhost/sakila'
sq --search-path customer1 '.customer_name'
sq --search-path customer1 '.customer_name'

...not sure what the equivalent to schemas/namespacing in other databases are so as to try and keep it as generic as possible.

[neilotoole]

@c-nv-s I agree. I'd like to implement both mechanisms. The flag is probably the lowest effort to implement. I've created a separate issue for it.

Thanks for the feedback, keep it coming!