You know the feeling of actually coding something, a new library which doesn’t exist yet but you know exactly how it should look and you just need to chip away at the problem space until its final form appears ?
I was really lacking any creative flow recently and decided to flesh out the libgnomesearch library that I proposed a while ago, which my colleague Kaspar Matas already made a start at developing.
The purpose of this library is to consolidate existing search code that currently lives in Nautilus, GTK, GNOME Shell and various core apps. There are some interesting designs for improving GNOME’s desktop search functionality, and I suspect it will be easier to prototype and implement these if there is a single place to make the changes.
I dedicated some evenings and weekends to the fun part of building an API and example app, and this is the result. Some code, some documentation, and a demo app:
To state the obvious : it’s not finished and it has bugs!
I can’t spend work time on this and probably won’t have much more volunteer time for it either, so let’s see what happens next. If this sounds like a project you are interested in continuing, get in touch – and if you are at GUADEC 2023 then we can chat there!
Sam Thursfield 
I really like this 🙂 A separate library / example app definitely seems like a good way to be working on this problem space. When you mention consolidating with Nautilus, I’m very curious what you picture there. Do you see a case for Nautilus making use of other search providers as well, or would we want to skip those and go straight for the built in Tracker search provider?
Hey Dylan! It’s a design question I suppose… I suspect it would be weird for Nautilus to return search results from the Characters app and such 🙂
What could be possible though, is for libgnomesearch to have both a Tracker and a “simple” filesystem search provider (or even an `mlocate`-based provider – crazy as that sounds). So some of the complexity that’s currently in Nautilus could move into the library. Just an idea