Hacker News Reader

Comment by blattimwind | Hacker News Reader

In SQLAlchemy it could look something like this

  cols = list(table.columns)
  cols.remove(table.columns['updated_at'])
  cols.remove(table.columns['inserted_at'])
  query = select(cols).select_from(table)

(You can abbreviate table.columns with table.c)

With sqlalchemy.orm you could also define a Bundle, see https://docs.sqlalchemy.org/en/latest/orm/query.html?highlig...

View on HN

Much simpler usint the ORM - you can defer columns:

http://docs.sqlalchemy.org/en/latest/orm/loading_columns.htm...

This is both at the 'object' level, and at the query level for example at the query level: from sqlalchemy.orm import defer, undefer

query = session.query(Book) query = query.options(defer('summary')) query = query.options(undefer('excerpt')) query.all()

You can also group columns for deferral/undeferral.

I use this loads.

Interesting, I knew about defer and used it in the past for pretty much the purpose the first example illustrates, but didn't remember/realize it could be used on a per-query basis.

I think if you strip columns for performance reasons for one specific query it would still be better to use Bundles or the equivalent loose select (i.e. session.query(Book.id, Book.title)) instead of query-local-deferred columns, because then you will notice if a code change leads to a down-stream access of an unwanted column for some reason.

what sql does this produce though? I can imagine that there'd be a hit on performance

Why would you imagine that? This compiles to SELECT col1,...,coln FROM table ...;

Reply on news.ycombinator.com