Database Q&A

[00:00:00]
>> Scott Moss: So this Drizzle studio is using our ORM. It's yeah, it's the same thing. Drizzle studio. Is a it's built by the same team, so it's from Drizzle Kit which is their COI which accompanies the Drizzle ORM.

[00:00:16]
Is a it's built by the same team, so it's from Drizzle Kit which is their COI which accompanies the Drizzle ORM. They're both one and the same, yeah, they work together. And those are the are the tables, it's literally reading our schema, right? So if I go to Here it'll show us our schema.

[00:00:33]
So if I go to Here it'll show us our schema. Where is it? Right here. Here's our schema, we define this.

[00:00:46]
Here's our schema, we define this. This is the sequel version of our schema. It's everything that we made, If I go into a Drizzle Runner, I get types like everything we just wrote in our seed script, I can write here, right? If I wanted to DB.

[00:01:03]
If I wanted to DB. Insert Entries. Values. I can make a new entry here.

[00:01:20]
I can make a new entry here. I can run this. Directly into the database. And then I guess here's the actual code that we wrote for our schema.

[00:01:33]
And then I guess here's the actual code that we wrote for our schema. It's a little more explicit because this actually adds the indexes like for instance, we did the shortcut.t unique on the end of the things that we wanted to be unique, but the long kind of version would actually look like this, so it just it did that instead, but yeah, this is the same schema that we wrote, this is the exact same code. Yes. Your experience, what's the best way to manage large amounts of seed data Do you just stick it all in seed.ts or do you have another solution?

[00:01:51]
Your experience, what's the best way to manage large amounts of seed data Do you just stick it all in seed.ts or do you have another solution? It depends, right? I mean, if we look at the notes, I kind of talk about it about a little bit. I think it depends on like the type of data you're trying to see, like, is it fake data like we just did where we're specifically Well I guess in our case it's a generic seed we're just like we're just generically Seeding the database we don't actually have like an intention or a use case that we're trying to solve.

[00:02:05]
I think it depends on like the type of data you're trying to see, like, is it fake data like we just did where we're specifically Well I guess in our case it's a generic seed we're just like we're just generically Seeding the database we don't actually have like an intention or a use case that we're trying to solve. So in that case, you know, yeah, you literally might have a directory where you just have like a bunch of fake data that you add, you know, that could be a JSON file that someone made somewhere that could be something that you generate with faker, however you wanna do that, but like in my experience, I typically. We'll have different scenarios that I wanna see the database with so I'll have like a folder called and like I'll have like a seed folder that I'll have like scenarios and each file is a different set of data for a different scenario that I'm trying to test for so I might have one scenario where it's like I am building a feature for when someone has more than 30 habits so I might see the database with one user who has 30 habits and nothing else so I can. Do that scenario because I'm building something on the UI that is addressing that we're building like pagination and maybe a table only has 10 rows so I wanna test out pagination against 3 pages so I see the database with 1 user and 30 habits so I can paginate against 3 pages right?

[00:02:22]
Do that scenario because I'm building something on the UI that is addressing that we're building like pagination and maybe a table only has 10 rows so I wanna test out pagination against 3 pages so I see the database with 1 user and 30 habits so I can paginate against 3 pages right? So I might have different scenarios that I'm sitting with so that's how I do it, but you know you very well might do it another way. You might just have JON files, you might. Again, do an export from a database and import into yours, if you're using something like we are Neon or Planet Scale, you can, branch off of a database and then therefore it already is seeded so depending on like the app and the company that you're working at, you know, working with production data.

[00:02:44]
Again, do an export from a database and import into yours, if you're using something like we are Neon or Planet Scale, you can, branch off of a database and then therefore it already is seeded so depending on like the app and the company that you're working at, you know, working with production data. On your, on your local machine may be OK, it may not be OK. Let's assume it is OK. Yeah, you could just like make a branch from production.

[00:03:03]
Yeah, you could just like make a branch from production. And get that URL attach it to your local machine and now you have a copy of the Production database and you have the same data that's in production you can. Now it's a real database and it's a real application so there really is no one way to do it There isn't like a best practice. It really is one of those things where it's like it depends what like what are you trying to do, but there's nothing wrong with just writing stuff the way that we did.

[00:03:16]
It really is one of those things where it's like it depends what like what are you trying to do, but there's nothing wrong with just writing stuff the way that we did. If that becomes hairy because you got too much stuff, then yeah, figure out, you know, what are some ways you can scale that out, but there there's no wrong way I think. This is actually like fake generating fake data is a really great use case for AI because it already hallucinates so bad. So like having AI generate seed stuff for you is actually a really great use case.

[00:03:32]
So like having AI generate seed stuff for you is actually a really great use case. Cool. Any other questions? Yeah.

[00:03:51]
Yeah. Is that little check we wrote at the end it was like ARGV 10 yeah, is that just to prevent. The file from being run directly outside of the terminal like exactly because like imagine if I imported this file from another file it'll immediately run. Oh, I don't, I don't want that.

[00:04:04]
Oh, I don't, I don't want that. I wanted to import it so I can call it once programmatically so this gives me the best of both worlds. This allows me to run it from the terminal and immediately execute And then this also saves me from it immediately executing it if I were to import this somewhere else, yeah. Cool.

[00:04:21]
Cool. Any other questions No. Cool, what else I have on here, so yeah, we got that. That was all there is to get the Database started now we do have.

[00:04:43]
That was all there is to get the Database started now we do have. These other commands here, the Generate and the migrate that we did not run. Don't worry, we'll run those when we get to production. So if you're wondering like, hey, we didn't do that, it's because we didn't need to.

[00:05:02]
So if you're wondering like, hey, we didn't do that, it's because we didn't need to. So the flow going forward is, let's say we have we have our schema, right? Let's say I wanted to, I don't know. Change something, do something.

[00:05:20]
Change something, do something. Right now, as long as I don't make a destructive change, I can keep doing push and I won't get any problems. I will run into a problem if I try to do a destructive change and try to do a push right? So if I try to, if I try to rename one of these fields that's required or something, and that data exists in the database already I'll get into trouble, but Because we're just pushing and because I see that this data anyway.

[00:05:45]
So if I try to, if I try to rename one of these fields that's required or something, and that data exists in the database already I'll get into trouble, but Because we're just pushing and because I see that this data anyway. I could just tell Drizzle to like yeah I know just blow out the data. I don't care so Drizzle just reset the database and push up my changes anyway that's not something you wanna do in production. You don't wanna just delete everybody's data and force push or schema you could literally go to jail, so like don't do stuff like that.

[00:05:57]
You don't wanna just delete everybody's data and force push or schema you could literally go to jail, so like don't do stuff like that. The way we get around that in production is with generating migration and then running the migration but for development. The flow you wanna do is like if you're if you're you know you're working on this PR and maybe you're making tons of changes to the schema or even just a little bit of change in the schema just keep pushing, just keep pushing and then once you get to the point you're like, alright, I'm done with this feature I'm ready to do a PR that's when you generate your migration because if you have to make a migration for every change that you do as you're figuring it out, let's say you made 6 migrations. You're gonna make a PR into production with 6 migration files.

[00:06:13]
You're gonna make a PR into production with 6 migration files. No one's gonna merge that. It's not gonna happen. You only want one migration file for all your changes versus 6 migrations for all 6 of your changes.

[00:06:30]
You only want one migration file for all your changes versus 6 migrations for all 6 of your changes. So use DBPush as a way to test your database until you're like, right, it's ready. I'm ready to do this migration now and when we get to deployment later today we'll