An Update on Ulry

If you’ve been following me lately you know that I’m working on a link archiver application that I heavily use to save articles and links that I can’t read when I see them. It’s an iOS only (for the moment) application that you can find at https://ulry.app and it’s currently available in beta test on TestFlight.

The app has been in the works for about a year now and it’s almost ready for v1.0 + App Store debut, but I must make some more changes to the database (:sigh:) and add some more features that I think are going to be useful for everyone that uses the app.

I don’t know how but I must have been drunk when I first created this database schema. And now it needs fixing…​ migrations…​ one thing out of place and all my fellow beta testers are going to write me agry emails saying that I messed up all their lovely links.

Right now I have almost 300 invites for the beta test application but only ~100 users are actively using it, which still is quite a lot of people enjoying my work and trusting that further updates are not going to cause any trouble, so I am taking my time.

I’m currently done with the adjustment migration with better indexes and triggers that are going to automatically handle database operations on certain events. I almost deleted 50 lines of code a couple of days ago after these triggers were added, so I’m happy with the results.

Okay okay, so… what else is missing from the final product?

Not that much actually! I would like to add full-text search and iCloud sync, but that’s probably going to be released in a future update.

I’m working on FTS at the very moment, it’s not that difficult because sqlite has a virtual table plugin (FTS4) that I’m leveraging to implement this very feature. It’s as simple as creating a new table

CREATE VIRTUAL TABLE search USING fts4 (id, url, title, description);
INSERT INTO search SELECT id, url, ogTitle, ogDescription FROM link;

Hmmm, why are you including that id in the search table? Isn’t it going to be indexed too?

Indeed, you’re right. At the moment I do not have a different way to lookup for resulting links using fts4. Query performances are still very good so I’m not bothering with this any further for the moment.

Maybe some caching logic would not hurt but for the moment performance is great, I can search everything I have directly from the app now and results are being returned in ms. I will try to test this out with a larger database but sqlite is too good to be slow when it’s under load by a single user, there’s no chance it’s going to slow down even with thousands of records.

The hardest feature to implement is going to be iCloud sync, it’s a totally new topic for me and I don’t even know where to start to be honest. Indeed, if you’re reading this and you have some cool ideas or you’ve done something similar in the past then email me straight away with some tips! To add more complexity, I have to create custom behaviours and logic because I’m not using CoreData, but a raw sqlite database. This means that I don’t get to use the built-in CoreData-CloudKit integration (:sigh:). Either way, this feature is going to come sooner or later, but I would love to release the first version of the app without it and consider iCloud sync when there are plenty of users onboard that find it a must-have.

I think I’ll make a small series about implementing the iCloud part with the help of ChatGPT, just for fun. Along with that I plan to write some technical articles about the not-so-complex behind the scenes of how Ulry retrieves and stores your favorite links.

Until then, see you on TestFlight and thanks for all the feedbacks!