News for hackers. Stuff that matters.
Re:SQLite 3.8.0 Release with next-generation query planner by orf on Tuesday August 27, 2013 @04:31AM (#6282190)
People in this thread say SQlite is fine for apps with lots concurrent writes (people advocate running forums and other multi-user apps using it) but that's not correct in my experience. I wrote a blog[1] in Python that uses a Sqlite database and on each post view a view counter was incremented[2] using a simple UPDATE ... views = views + 1 on the post row. Removing this update caused page speeds to increase hugely, from ~10 requests a second to ~200 or so. Maybe I was doing it wrong, but I can't see how.

1. https://github.com/orf/simple

2. https://github.com/orf/simple/commit/2121401030cab4d500330e9...

Re:SQLite 3.8.0 Release with next-generation query planner by sk5t on Tuesday August 27, 2013 @11:50AM (#6284178)
I'm not familiar with the ORM here, but is it really just an increment ("update posts set ctr = ctr + 1 where id = 9") or a select, add one in code, and then an update? Also bear in mind that one operation in an explicit transaction like this is quite slow, at least prior to WAL.