The tagging system is indeed flat, but the lists can be nested. The idea being that tags are usually AI generated, and there's a lot of them (which is useful for search), but lists are meant for manual curation and this is where you can have whatever structure you want.
IMO it would be interesting to try to combine the two approaches (curation + auto tagging).
It starts out with the user scaffolding an initial hierarchy, then (after enough usage to provide meaningful data for ML predictions) the ML model predicts on subsequent entries, and asks the user for approval (which feeds a reinforcement learning model)
This is indeed the plan. We're currently working on generating embeddings for the all the bookmarks stored, and one of the usecases of this is going to be clustering. If a bookmark is similar to all other bookmarks in a list, the model can suggest adding those bookmarks to the list. Still a manual operation, but with ML assistance.
Hoarder supports customizing the tagging prompt in the user settings. You can instruct the model with whatever rules you want to constrain the tags to your likings!
I must ask, in your github readme you say "[Planned] Downloading the content for offline reading." but what does that mean in the context of a self hosted application? Isn't the data already downloaded - via the other features such as "Full page archival" and "Automatic fetching".
I guess one small request - could the chrome/firefox extension include a way to transfer the page data from the browser, as it's being displayed to the user? (as in, transfer the entire page/html instead of the page's link). This would likely result in much better support for nasty sites like twitter and such that require credentials, etc..!
The offline reading thing is bad wording from my side. I mean offline reading on the mobile app (when you don't have internet access / access to the server).
FYI: On the login, it tripped me up a couple of times because the username is case sensitive. There is a tradeoff between security, useability, and support requests; the input is labelled email, and email addresses are usually not case sensitive (and as email addresses used as email addresses are never case sensitive) so it confused me.
TiL. I always assumed emails were case sensitive, and doubly so if used as a username. I find it strange that you even discovered this 'wrong' behaviour on the site in question: you purposefully typed your email address with different casing when logging in vs. registering?
Although the standard requires the local-part to be case-sensitive,[1] it also urges that receiving hosts deliver messages in a case-independent manner,[2] e.g., that the mail system in the domain example.com treat John.Smith as equivalent to john.smith; some mail systems even treat them as equivalent to johnsmith.[3]
You'll find the footnote links at the Wikipedia article, I'm not going to paste the here.
So yes, if my email address is my username, then I would expect it to work the same in uppercase or lowercase. If my username is "like an email address, but not an email address" then you make the rules for your site.
Question for ya since I'm working on an app with similar platforms - and dont know what I'm doing. I see you used expo for the mobile apps but nextjs for the web app. Why didnt you use expo for the web app too?
Hey, Hoarder's maintainer here. I'd like to know more about the pain you're facing when upgrading hoarder. All the releases since launch has been backward compatible, and it has always been just a matter of updating the docker images.
Also for searches, Hoarder indexes all the content of the websites it crawls. If it doesn't for you then that would be a bug!
Thanks for the advice will do that for sure! Concerning your questions I applied mostly to US based companies( Facebook, Google, Quora, Dropbox, Box, Pinterest, ..).
