Hacker News Reader

Comment by reallymental | Hacker News Reader

Incredible, but wasn't this kind of available though psycopg2 (postgres connector), SQLAlchemy or any other kind of database connection library ?

I realise the difference between the kernels altogether, but how is one better than the other?

View on HN

Weird as it might sound, I can imagine a scenario where I wouldn't need anything past SQL in some form. I've got plenty of python scripts sitting around my hard drive that are just fronts for data manipulation. Especially with the cvs import mode, I could see myself using just sqlite or whatever

PSA: SQLite is embedded as a standard builtin python library.

no pip install anything just import sqlite

https://docs.python.org/3.7/library/sqlite3.html

But I will be trying the Jupyter kernel in the article

Not really. You could use the python kernel and embed your SQL code within your python code but you couldn't run cells with just SQL. It made for a very poor analyst experience.

Microsoft got it right with Azure Data Studio.

Does ADS support SQLite? I already use it for SQL Server, and I'd tried the PostgreSQL add-on but couldn't get it to work for some reason (but that was when it was first released so it might've stabilized since then).

No. I've seen success with Microsoft SQL and PostgreSQL.

I've also experienced instability issues, it needs that VS Code level polish.

Yes and when paired with Pandas `read_sql`[0] it's easy to quickly query, view and plot.

There's also pandasql[1] which is based on SQLite. We use this behind the scenes at SeekWell[2] to power cross database / cross source joins.

0 - https://pandas.pydata.org/pandas-docs/stable/reference/api/p...

1 - https://pypi.org/project/pandasql/

2 - https://seekwell.io/

Well, as someone who strongly dislikes Python and strongly likes SQLite, this might give me a reason to willingly use Jupyter (granted, I guess I could do it through Julia, too, which I also strongly like, but... eh).

A lot of the time when I need to work with a bunch of data I'll fire up the SQLite CLI, snarf the data with the built-in CSV import commands, and do whatever queries I wanna do. This looks like it can provide a nicer UI around that workflow, and I'm all for it.

Datasette looks interesting for this, too; I'll have to give that a look.

I keep a SQLite buffer open in emacs, might switch to this given it's nice visualizations.

Reply on news.ycombinator.com