text stringlengths 0 2.29k |
|---|
**Gerhard Lazu:** Exactly. Yeah, he connected a couple of dots. Also, I've been running them long enough to understand the trade-offs which I'm making, so it's a combination of things. |
**Lars Wikman:** That's the tricky bit, I think... Because to me, if something is implemented in an Elixir app, so the app sort of manages itself - that's easy to read for me, that's easy to reason about. I know how Elixir apps work. Bash scripts, I'm okay with; makefiles, I have a hard time reading. They don't flow li... |
\[42:25\] When I was in the Python ecosystem, it was like "Oh, I want to talk to a server." Well, Ansible is actually written in Python, so there's some synergies there. I don't get super-confused when some Ansible package breaks. It's like, "Yeah, yeah, Pip... Let's Pip. Just Pip things." And similarly, when I need to... |
And if you're doing Go and you're not into-cloud native, I don't know what you do, because it seems like there's tons of tooling for everything written in Go, but it's all for the cloud-native space. But I think staying pretty close to whatever culture you're in, or whatever your comfort zone is, is a decent way of kee... |
**Gerhard Lazu:** Interesting. Do you remember the name? |
**Lars Wikman:** I think the project is called CI, under his GitHub, but I don't think it has continued. I think he just ran out of bandwidth, and it potentially became apparent as well... |
**Gerhard Lazu:** Yeah. That changes things a lot, for sure. Okay. |
**Lars Wikman:** That's actually one of the reasons I got curious about what you were saying about Dagger, and having sort of this base support for building SDKs for different languages... Because I would like to write Elixir for my CI/CD. |
**Gerhard Lazu:** Interesting. Okay. |
**Lars Wikman:** Because I'm very proficient in Elixir, so... |
**Gerhard Lazu:** Yeah. I think there's something really interesting there, because you're right, you need to have an interface that you're comfortable with, and I don't think YAML is it. I don't think YAML -- well, I mean, we just make do with it, and that requires... It really is a requirement, a declarative system. ... |
So YAML - great for declaring a state of the world, but then there's all sorts of transformations that need to happen, all sorts of functions need to be called at different points in time. That YAML basically has to be reconciled into something useful, which is what Kubernetes is, in a nutshell. Okay, I'm oversimplifyi... |
**Lars Wikman:** And if my job was managing a complex system day in, day out, and not mostly developing the system, if I could spend most of my time on the operations part, then Kubernetes might also make more sense, because then that's a tool that gives me a lot of capabilities, and I can spend my time learning to be ... |
**Gerhard Lazu:** UCP. |
**Lars Wikman:** \[46:19\] It all boils down to what I'm comfortable with. Like, I've done Linux since I was a teenager, so I know how to do Linux. |
**Gerhard Lazu:** What about systemd? Are you okay with systemd? |
**Lars Wikman:** I'm getting okay with systemd. |
**Gerhard Lazu:** \[laughs\] I know... That's such a hard thing... runit, please. Can I get runit back? That was my favorite supervisor. It was so simple. That was like the pinnacle of supervisors for Linux systems. And then systemd came along. |
**Lars Wikman:** It seems very capable, I'll say that. And perhaps in ways that would be hard to replicate with like in .d files, and scripting all that on your own. It seems very capable. |
**Gerhard Lazu:** My systemd is your Kubernetes. Not gonna happen though... \[laughs\] |
**Lars Wikman:** Yeah... But, I mean, systemd under Kubernetes makes very little sense, I think. |
**Gerhard Lazu:** Yep. |
**Lars Wikman:** It's like, no, your containers do not need systemd. |
**Gerhard Lazu:** And there you have it. I wanted to avoid systemd so badly, that I switched to Kubernetes... \[laughs\] Because I was shocked by all the horrors that would happen in systemd. And good luck figuring out those units... |
**Lars Wikman:** You wanted systemd at a multimachine scale. That's what you wanted. |
**Gerhard Lazu:** Exactly, yeah. Exactly. |
**Lars Wikman:** Have you poked around with other non-Linux operating systems, like the BSDs, and things? |
**Gerhard Lazu:** Yeah... I used to run FreeBSD for the best part of the last decade. It's interesting. Jails were interesting. Solaris zones - I only worked on a project for maybe three or six months that was using it. They seemed very complex, solaris zones, like from the outside. There was like a lot of stuff that w... |
To be honest, I understand the appeal of using something that you're comfortable with. Something that you're like on a trajectory, and you've been on that trajectory for a really long time. You mentioned Linux - it does most of what you need. Of course, some parts are not perfect, and you're not happy with, but is ther... |
**Lars Wikman:** Yeah. And for me, that's sort of not really a question, because I always explore new things. I don't really ever fix, like "This is how I do things." I do that for a project, for a time, like "Okay, this is how WE do things." That doesn't mean that's how I do things. I do those things in that context. ... |
One reason I asked about the BSDs is I've gotten some really good, fun input about operational stuff from one person that reads my newsletter, and he works on FreeBSD, so he contributes to FreeBSD, and I think he works on CouchDB as well. It might be someone you should have on the show, DCH Dave Cottlehuber. I hope I s... |
**Gerhard Lazu:** \[50:06\] Interesting. |
**Lars Wikman:** So that could be an interesting conversation... |
**Gerhard Lazu:** Interesting, yeah. |
**Lars Wikman:** And I know he's super-comfortable with all of the networking that I rarely touch; he's done sort of operations at a scale I've never had to. And he chooses FreeBSD. I think one of the reasons is that it's simpler in many ways than Linux. I get that impression from people that choose the BSDs, that it's... |
**Gerhard Lazu:** Yep. CSH - Oh, my goodness me. I mean, that thing is just -- like the shell, the default shell on the FreeBSD systems that I was using... And it's just behaving in an unexpected way for someone that's familiar with Bash, or ZSH, or even Fish. |
**Lars Wikman:** Yeah. I think going off the beaten path is good and useful, but I think it also adds up... So if you spend all of your time and effort way off the beaten path, you're gonna get yourself in trouble. Or you're gonna build absolutely brilliant systems that no one else can work with. |
**Gerhard Lazu:** That's the problem, isn't it? Works of art, as you mentioned. It's a work of art. It's amazing, but no one knows about |
it. NixOS, and PureScript, and Haskell all the way... |
**Gerhard Lazu:** Yup. OCaml, don't forget about that. And a couple others. Zig... |
**Lars Wikman:** It's like Fortran, where it needs to go fast... |
**Gerhard Lazu:** Yeah. With a bit of COBOL to keep everything together. As we prepare to wrap this up, Lars, any one takeaway that you would like our listeners to have from our conversation? |
**Lars Wikman:** I think don't worry too much about sort of the popular tooling right now. Use something that you know that you can make work, and mind how many new things you introduce... Because whenever you bring in new things, you are challenging yourself, and you probably should challenge yourself on a regular bas... |
**Gerhard Lazu:** Excellent. Well, thank you very much for today. This was good. There's one more year that goes by, that we have similar conversations... And I'm very curious to see what happens next time. |
**Lars Wikman:** Let's see if I change... |
**Gerhard Lazu:** Yeah, exactly. Or if I change. It has happened... Until next time, Lars. Thank you. |
**Lars Wikman:** Thanks for having me. |
• Gerhard Lazu is leaving Ship It after 90 episodes |
• The decision is due to burnout and needing headspace for other projects |
• Change is constant and embracing it is key to moving forward |
• Gerhard wants to focus on experimentation and improvement in the future |
• The possibility of reviving Kaizen-style shows or podcasts in some form is discussed |
• Discussing a podcast's new format and infrastructure |
• Reflecting on past shows and their successes |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.