Monday, 21st May 2018
The big thing I always get asked to find are dank dilapidated alleys, and New York City has, like, 5 alleys that look like that. Maybe four. You can’t film in three of them. So what it comes down to is there’s one alley left in New York, Cortlandt Alley, that everybody films in because it’s the last place. I try to stress to these directors in a polite way that New York is not a city of alleys. Boston is a city of alleys. Philadelphia has alleys. I don’t know anyone who uses the ‘old alleyway shortcut’ to go home. It doesn’t exist here. But that’s the movie you see.
VirtualKNN for SpatiaLite. This looks amazing: a special virtual table shipped as part of SpatiaLite 4.4.0 which implements a fast, R-Tree backed mechanism for finding the X nearest points against a geospatial database table. There’s just one catch: it’s only available in 4.4.0, but the most recent “stable” release of SpatiaLite is 4.3.0a from September 2015 so the version you get if you install from apt-get or homebrew doesn’t yet have this functionality. I’d love to figure out a neat way to package and distribute this along with Datasette. I’d also like to figure out a clean way to ship a more recent version of SQLite than the one that is currently packaged with Python 3 (3.16.2, where the latest SQLite release is 3.23.1).
New in Django 2.0: Database instrumentation. I missed this previously. Django 2.0 shipped with one of my most-wanted features: the ability to easily instrument database calls (for logging and metrics) without having to monkey-patch or run an entirely new database backend. Can’t wait to try this out.