On: SpatiaLite: Extends SQLite core to support Spatial...

[zipzappy]

It’s not actively developed. The nearest neighbor functionality in the released version is deprecated and both it and the development replacement have a recently filed bug that makes you worry about accuracy. The docs are also out of date

[teruakohatu → zipzappy]

This. I strongly recommend not using it at this point in time.

SQlite really lends itself to having a spatial backend and would likely become a bit player in this space if it did, but it probably needs a major overall first.

[fortysixdegrees → teruakohatu]

Strongly recommending not using it in a blanket way is not helpful.

I have used it extensively in an embedded product for many years, and it has just worked.

Sure it doesn't have all the postgis goodies, but it has enough that it's not usually hard to find a good way to do the spatial queries you need.

I'm quite surprised by this thread so far, given how robust it's been for me for many years using it in many different ways.

[teruakohatu → fortysixdegrees]

To be specific, they changed the KNN backend, but the docs refer to the new unreleased backend. The old one is deprecated but unreleased, which is a strange confusing situation.

https://www.gaia-gis.it/fossil/libspatialite/wiki?name=KNN2

[fortysixdegrees → teruakohatu]

For the average HN reader spatialite is a tool which should be considered. If the KNN issues are a breaking point for someone, they are probably already in postgis territory.

[fortysixdegrees → zipzappy]

It is immensely useful as is with no active development, although I'll have to take your word for that, I wasn't aware it isn't active - that has never affected me after years of heavy use