text stringlengths 0 1.82k |
|---|
**Brian Ketelsen:** It is. |
**Ben Johnson:** Computers are hard. |
**Erik St. Martin:** I think we got 10 or 15 minutes left; did you guys want to talk about anything else going on in the community or any interesting projects? I know I mentioned the noms database, I don't know if you've got a chance to look at that. |
**Ben Johnson:** \[40:04\] No, I haven't yet. Is that the one based on Git? |
**Erik St. Martin:** Yeah, that's exactly the one. |
**Ben Johnson:** Okay, no, I haven't taken a look yet. |
**Erik St. Martin:** So I haven't played with it, but I thought it was pretty interesting because one of the things that appealed to me was… It has the content addressable, which is becoming popular too, in storage, and the append-only, so you only get versioning. But one of the things why I really want to play with th... |
**Brian Ketelsen:** That gives me nightmares, thinking about the database rebase \[laughter\]. I am not rebasing my database. |
**Erik St. Martin:** So that becomes the question, right? I have not played with it. Don't take that as me advocating that you should go out and try and use this for that. But it did trigger that idea that I'd like to play with it, and see how well that works. |
**Brian Ketelsen:** I'm curious to take a look at it. I didn't have time to read the readme before we talked, but the tradeoffs that they've chosen are interesting, because in databases everything is a tradeoff. You have to choose. So what's the use case that they're trying to solve? |
**Ben Johnson:** Yeah, I was thinking that too, I wonder what their use case was. |
**Erik St. Martin:** They have a Slack, a mailing list, and a Twitter. We should ask them. |
**Brian Ketelsen:** Yup. Well, they do, they've got it all. They even have a cute logo. |
**Ben Johnson:** Typically, we mention these things, and then somebody who works on it, or knows a person who works on it just happens to be in the Slack channel, and we're being fed information real-time \[laughs\]. |
**Brian Ketelsen:** It's not happening today. Why are the noms people not in GoTime FM Slack? Somebody fix this. |
**Carlisia Thompson:** Talking about cute logos, the project I want to mention today also has to do with database and it totally won me just for the website, because the website and the logo is also cute and well put together. It's simple and clean. So if you want to sell something to me, you know how to do it \[laught... |
**Erik St. Martin:** Give it a cute logo? |
**Carlisia Thompson:** If it's cute and color-coded, there we go. I'm sold. |
**Erik St. Martin:** I have this blade of grass, but it comes with a very cool logo. \[laughter\] |
**Carlisia Thompson:** Exactly. |
**Brian Ketelsen:** Sold! |
**Carlisia Thompson:** I'm easy like that. So this is not an ORM, this is a productive data access layer for Go. I'm finding this so funny because I think in every other language, a package like this will say, "We are an ORM. We are the best ORM", no matter how full-featured it was. But with Go, it was like, "No, no, n... |
**Brian Ketelsen:** It's a dirty word in Go. |
**Carlisia Thompson:** It's a totally dirty word, right? So much. I chuckle every time. |
So I was looking at not ORMs and ORMs recently, and I found this today - it was actually in the Go newsletter, and I think I'm going to use it. It's not full-featured, it's not doing a lot of things that makes me scared. Being very new to Go, I don't think I can recognize enough to make decisions; my package is doing w... |
**Brian Ketelsen:** \[44:20\] That's where people in the programming world always differ. There's a category of people that want to write their SQL perfectly tuned by hand, execute it, and then map those back into structures, and use them for whatever they need to. There's another category of people that doesn't want t... |
**Carlisia Thompson:** Yeah. |
**Erik St. Martin:** So I'm actually somewhere in the middle. |
**Brian Ketelsen:** Oh, you just had to be contrary. |
**Carlisia Thompson:** With this, you can totally just send SQL. So I like that because I anticipated that I would need to do that. If it didn't have that feature, I wouldn't use it. But I think all of these libraries will have that, I would imagine. So you get something... You get both things. |
**Erik St. Martin:** Some of my pain points with doing straight SQL is… Anybody who's tried using the standard database SQL package can attest to this - scanning individual things gets painful. Especially when you're changing the SQL, you didn't get your type right, and all these things. That can cause a lot of pain, s... |
But there's also the other side of it. We talked about Rails. I love how easy Rails is to do these things, but there's two sides of that equation. It's not saying that I don't believe in that, because I really do; that fast prototyping is awesome. But there's two issues I end up having occasionally with it. One, when y... |
There's this translation process, but for the most part if you get familiar with Active Record, that's not a big thing. But for new people, there's also the fact that things like major ORMs can be leaky abstractions. They're perfect abstractions when everything is going well, but if you don't understand the SQL that's ... |
**Carlisia Thompson:** Yeah. There is the performance issue that if you're not aware what's going behind the scenes, you can get into trouble. That might also be the issue where you want to do something a little bit different, and now what? But you can always fall back to SQL, I guess. |
**Erik St. Martin:** It sounded like, from your standpoint, Ben, that you like the ORM side of things; you think that's a point where we could be better at, that would grow Go usage? Adoption, rather. |
**Ben Johnson:** I wouldn't even necessarily say that. I think when you get into ORMs, unless you're generating the code… A lot of ORMs use a lot of interfaces, and you lose the type-safety, and you just get a lot of issues on that side. I think there's some fundamental issues from mapping objects to relational ideas t... |
\[48:07\] So I wouldn't say ORMs are the way to go. Personally, I would avoid SQL, if you can. But that's my own personal opinion. |
**Erik St. Martin:** All key/value store, all the way. |
**Ben Johnson:** All the way. |
**Brian Ketelsen:** If you were going to avoid SQL, how would you build a system that had to have more than one active copy of the database? Is it possible to do something like that with Bolt? |
**Ben Johnson:** More than one active copy... Like distributing it in a cluster? |
**Brian Ketelsen:** Yeah. |
**Ben Johnson:** I think you can do things around. I think you have the same tradeoffs, whether it's Bolt or some other database where if you need to scale things horizontally, typically you need to shard your data. If you have a lot of coordination inside a distributed system, even a database, there's going to be lock... |
**Brian Ketelsen:** Like a standby? |
**Ben Johnson:** Yeah, a standby. That's one alternative. |
**Brian Ketelsen:** So does BoltDB offer any sort of replication log now? Synchronous or asynchronous? |
**Ben Johnson:** Not currently, but that is something I want to include in there. |
**Brian Ketelsen:** Nice. |
**Ben Johnson:** Because I think that would solve probably 95% of these cases out there, with a database and a standby that you can fall over to. That's why a lot of people run with Postgres, that uses async logs by default, I believe. |
**Erik St. Martin:** Right. So I think we're starting to close down on time, and I want to make sure we have time to go through Free Software Friday, but there's one other project you've been working on that I want to take the time to mention and that's the Secret Lives of Data. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.