Weekly Recap #4

[kaspth]

Hey everybody, we're almost a month in now!

I had a call with @swanson and @marcelolx last week, and the existing code we had was great at giving the discussion a lot more dexterity. So we could fairly quickly go yay or nay on what we had.

We ended up with some fairly big changes too, some of these are things we've discussed a bit before.

• We're further splitting the 1-1 between a fixture file and a table, and now we're starting a convention of grouping seeds around a top-level domain object like an Account/Team/Organization ala test/seeds/accounts/business.rb. This also means our seeds file are more like config/routes.rb then, they take up more lines but are grouped in one file — we can always look at better abstractions later. I've already converted this work in Group seeds by top level model + plus data tables #28
• Following on from Group seeds by top level model + plus data tables #28, I've also added a test/seeds/data folder for any pure data tables folks expect to have in their db.
• This new convention should also make it way easier to work on file checksum based caching, e.g. if nothing changed we can skip the seed. I've done some sketches on this, but I'm still figuring out a v1.
• We're making ids optional, so users.update name: "Kasper" is possible. This needs a little more work because we now need an alternate unique attribute to find by. I'm doing this in Make ids optionals #32

I'm also working on getting us to be able to use Oaken via a package provider of some sorts. I've kept deferring it since I knew it would be kinda pointless without making the changes above. Soon though!

PRs & Discussions

• @tcannonfodder shared ideas in Ideas for data reuse, Oaken use cases #31
• I've started setting up testing with multiple database adapters in Local test support for different database adapters #29
• I've added some tests for helper methods in Test helper methods #30, I still don't know quite what I want these to be able to do, but I just wanted these tested
• Converter: refactor + support diffing & namespaced models #27 I refactored our fixture converter a bit so we get Rails generators' (through Thor) built-in diffing for generated files

Next Team Call

I've just put up the meeting details over here #33

---

How's it going so far for everyone? Is it still manageable to follow along? Are y'all still learning Ruby tricks from seeing the code now evolving in a slightly different direction?

[BakiVernes]

Hi Kasper,

Can you please expand a bit more on this:

We're making ids optional, so users.update name: "Kasper" is possible. This needs a little more work because we now need an alternate unique attribute to find by. I'm doing this in #32

[kaspth]

@BakiVernes sure yeah,

So right now we've got users.update :kasper, name: "Kasper" where :kasper gets hashed and becomes the id of the created record, in the exact same way that fixtures behave:

kasper:
  name: "Kasper"

While fixtures require that everything added gets a label/id, we don't always need that. So I'm trying to add support for just users.update name: "Kasper", but then we need a different column or columns to compare uniqueness against. Otherwise the internal find_by call that update does can't find anything to update which breaks the conceptual model we're working on.

Let me know if there's something that's still not clear!

[BakiVernes]

So in the PR that you mentioned are you gonna tackle the missing find_by column?

[kaspth]

@BakiVernes yes, exactly. I'm still trying to figure out how that's gonna work. I'm also going to be testing to see if our default attributes through with could work, ala users.with id: -> { name }.

[manuelfcreis]

An interesting tweet by Nate Berkopec I saw today 😅

https://twitter.com/nateberkopec/status/1570806217546555401