Bug Report: Under some conditions NULL isn't NULL.
I was working on a project for an online course I'm taking and ran into a problem where NULL values resulting from missing rows in a LEFT JOIN weren't being treated as NULL values.
It took me a while to conclude the problem was with SQLite, rather than my code. But I compared the behavior I was seeing locally using 3.33.0 with a cloud-hosted version using 3.21.0 and found that I couldn't reproduce the behavior with the earlier version.
I've done my best to show the problem, but I'm also near the limits of my present knowledge and skill, so I hope it is enough for someone else to run with, or let me know that it behaves as it should.
I've shared a static rendering of jupyter notebook illustrating the problem.
I've also made available a zip file of the notebook and database I was working with.
Thanks! I'll have to wait until it works its way through releases and packaging. In the meantime I think I found another way to get at what I need.
(4) By sean (jungleboogie) on 2020-08-20 04:08:29 in reply to 3 [link] [source]
See this post about compiling sqlite: https://jvns.ca/blog/2019/10/28/sqlite-is-really-easy-to-compile/
Thanks. That's a useful link.
It's more a matter of time and attention. I already feel like my python environment is a house of cards, despite my efforts to make it otherwise. Adding a custom build of a library would be one more thing I don't feel on top of.
FWIW, the SQL from that Query-Flatting doc you link to looks weird.
I hesitate to write incorrect, since I'm no expert, but
AND="1" look suspicious to me. Wasn't the LHS of the = omitted?