Adds backpacks!

[benbot]

• Tested that that the server runs
• Tested that everything still works

This PR adds backpacks!

These are like inventories where users can keep collections of maps, avatars, and props made by other users.

The intent is to use these collections of items to feed a recommendation algorithm.

I also think it's a cool idea for players to have "starter packs" of props/maps/avatars/etc. that they can share around. This sets the foundation for that.

[fire]

Is there functionality to add, duplicate and remove backpacks? Add backpacks to backpacks? What about roles and access to the backpacks?

Edited:

Ok to be in the next pr.

[benbot]

Yeah, pretty much all of the features need to be added in, but the data model is weird enough that i thought getting a PR open for it first would be good.

I'm adding some very simple crud views for this, but as discussed, I'm going to try avoiding next and just use regular ol' templates

[fire]

Pretty please some unit tests :3

[dragonhunt02]

I think we do need a gated API endpoint in backend to properly use this feature

[benbot]

I should clarify, this just adds the schemas and a migration.

No new features were added.

Because of the way user content works (via some macros) it was a little tricky to set up the different relationships needed.

I think it'd be good to get an OK on this data model before I try building anything else on top of it, so that's really what I'd like some feedback on here.

We can hold off on merging this until I have something built on top if ya'll want, but i'd still like some feedback on this

[benbot]

I'm adding some simple crud views.

When those are finished up it'll be easier to test.

[fire]

With the recommendation algorithm from https://librecommender.readthedocs.io/en/latest/serving_guide/rust.html as long as we have user_id, item_id and label (0 or 1 for ranking). We're good.

In the future we can add jsonld attributes on either user or item but it's later.

For the future my ideal algorithm is https://librecommender.readthedocs.io/en/latest/api/algorithms/din.html.

I'm undecided if we want our labels to be 5 stars or merely is binary of a bag or not.

[dragonhunt02]

Because of the way user content works (via some macros) it was a little tricky to set up the different relationships needed.

About it, I think it was done this way to simulate inheritance in elixir

I think it'd be good to get an OK on this data model before I try building anything else on top of it, so that's really what I'd like some feedback on here.
We can hold off on merging this until I have something built on top if ya'll want, but i'd still like some feedback on this

I see 👍

[benbot]

About it, I think it was done this way to simulate inheritance in elixir

yeah for sure. I think using embedded_schemas would be the idiomatic way to have something like inheritence, nowadays.

... also now that I think about it, maybe I can just have one big join table for all the different kinds of user content, instead of making a join table per user content table.

[fire]

whats missing in this pr?

[benbot]

• Some tests to ensure that the relations are set up how i expect
• See if i can have one join table for all user_content, instead of a table per user_content type

[fire]

Will we have more content types?

[benbot]

@dragonhunt02 feel free to leave a review too btw. I would've added you as a reviewer as well, but you weren't showing up in the list 😅

[benbot]

Will we have more content types?

I can only assume we would.

[fire]

Is this ready to merge?

[benbot]

@fire its done from my end, unless anyone has comments