text stringlengths 0 2.35k |
|---|
• Some Go developers are excited about the possibilities of eBPF and are working on related projects |
• The past tense of "screenshot" could be "screenshat" |
• Optional second return arguments in Go should not be optional and require explicit use or omission |
• Certain generic abstractions, such as input/output and source/exporter pairs, may indicate overly abstracted code. |
• Over-generic abstractions can lead to complexity and inefficiency in software design |
• Specificity and opinionatedness are generally better than over-generality when it comes to abstraction |
• Standard libraries can be inefficient due to their need to check for edge cases |
• YAML is often used as a configuration format, but can lead to complexity and limitations |
• Go's execution tracer is often misused and can be counterproductive in debugging efforts |
• Go should provide more control over memory allocation for certain use cases |
• Rust-style ownership model could be useful in certain parts of Go code for strict control over variable lifetimes |
• The Go team should be an independent third party, rather than controlled by Google alone. |
• Discussion about having representation from the Go team to adjust things |
• Mat Ryer's reluctance to discuss controversy and suggestion to move it to Twitter |
• Mention of future episode topics, including profiling |
• Conversation wrapping up, thanking guests and listeners |
• Discussion of edited podcast version vs live YouTube video |
**Mat Ryer:** Hello, and welcome to Go Time. I'm Mat Ryer, and today we're talking about making Go more efficient, and how best do we do that. I'm joined by Bartek Płotka, author of Efficient Go, which was released last month, principal engineer at Red Hat, and a Prometheus and Thanos maintainer. Hey, Bartek. Welcome b... |
**Bartlomiej Płotka:** Hello. Hello, everyone. How are you? |
**Mat Ryer:** Very good. Thank you very much. And you're calling in from London, right? |
**Bartlomiej Płotka:** That's correct. Yeah. |
**Mat Ryer:** Yeah. Cool. I'm also joined by Kemal Akkoyun. |
**Kemal Akkoyun:** Hello, Matt. |
**Mat Ryer:** Hello. You're a software engineer at Polar Signals, right? And also a Prometheus and Thanos maintainer, and you're currently building an open source eBPF-based continuous profiling solution called Parka. Welcome. |
**Kemal Akkoyun:** Hello. |
**Mat Ryer:** Great to see you. This is your first time on GitHub, I think, isn't it? |
**Kemal Akkoyun:** Yeah, it is. |
**Mat Ryer:** Yeah. Good. |
**Kemal Akkoyun:** A bit nervous. |
**Mat Ryer:** Yeah. Oh, don't be. I am a bit as well, just because of the sheer brainpower on this episode... I'm out of my depth. I'm also joined by Christian Simon. Christian is a software engineer at Grafana Labs, working on observability databases like Loki, Mimir, and the new one, Flare. Hello, Christian. |
**Christian Simon:** Yeah. Hi there, Mat. How are you? |
**Mat Ryer:** I'm good, thanks. Welcome to Go Time. It's nice to have you here. Let's get to know our guests a little bit, shall we? Kemal , what's your favorite drink? |
**Kemal Akkoyun:** I guess I have two... One is single malt scotch, and the other one is coffee. |
**Mat Ryer:** Oh, nice. Yeah. Do you have them together? there's the Irish coffee that we call it, if we mix them together... |
**Kemal Akkoyun:** No. I have them separately, one in the morning and one in the afternoon. |
**Mat Ryer:** Okay, we won't ask which is which... What about you, Bartek? That sounds pretty good, doesn't it? |
**Bartlomiej Płotka:** Yeah, definitely. That's a +1 from my side as well, coffee and whiskey... And I got approval from my personal trainer as well. He said "Whiskey... Is it clean at least?" and I was like "Yep, clean. Just ice" and she was like "Okay, that's fine." So I think it's good. |
**Mat Ryer:** Oh, wow. Yeah, that sounds like a good personal trainer... Christian, do you have a favorite tipple? |
**Christian Simon:** Yeah. I think when I'm at home, I'm definitely more on the juice surely, \[unintelligible 00:05:10.08\] basically, as I'm originally from Germany, so that's the drink there... |
**Mat Ryer:** I see. |
**Christian Simon:** But yeah, I think given the feedback from the personal trainer from Bartek, I think I might need to change that and revisit. |
**Mat Ryer:** Yeah, no wonder you're not ripped... You're not drinking enough whiskey. Oh, you're from Germany, so -- okay, that's why we're doing an episode on making Go more efficient. Makes sense. That's the classic stereotype, everybody, which I don't recommend... |
**Christian Simon:** Yes. People haven't lived in Germany. |
**Mat Ryer:** \[laughs\] I don't know... Well, you have, haven't you, Kemal? Because you live there now. Do you like it? |
**Kemal Akkoyun:** I really like it. I like Berlin. It's been five years... Learning to be a German. |
**Mat Ryer:** Ja... |
**Kemal Akkoyun:** Slowly. |
**Mat Ryer:** \[05:58\] Nice. I think Berlin is a great city as well. GopherCon EU was there this year, so I got to go and hang out in Berlin, and we had a great time. That's the small talk over; the small talk section is now finished, okay? No more small talk now. Let's get down to business. I want to know just someth... |
**Bartlomiej Płotka:** So there are two schools actually, right? One is saying that premature optimization is the root of all evil; it's like a very old saying that got very, very, very popular, especially in the '80s and '90s... And that was kind of claiming that engineers are playing too much. They're having too much... |
So we can go deeper, but generally, yes and no. Generally, especially if you're learning, focus on readability, maintainability, and those kinds of first things that -- you have to learn how to make your code functionally running, so kind of like fulfilling your functional requirements, and then you can optimize it lat... |
**Christian Simon:** I would definitely agree. I guess, as with many other things, I think a balance of the two is probably good, and kind of the better you understand your ability, your team's abilities, but also maybe the problem space... Like, let's say you attempt to rewrite something you think you understand quite... |
**Kemal Akkoyun:** Yeah, I totally agree for both of the statements... But I guess as Bartek put that, I belong to the first school. The original quote is from Donald Knuth, and it says 97% of the time. So I think that 3% sometimes makes the difference. And you can always determine that 3% by measuring things. In Go we... |
**Mat Ryer:** \[09:45\] Yeah, so this is interesting then, and I kind of like what we're going to get into... Like, this idea that we don't really know what we can do to make it more efficient necessarily. And sometimes, even experienced engineers will have a really good idea, it'll be really obvious, but it turns out ... |
**Bartlomiej Płotka:** Well, we need to make some measurements. We need to observe our system under load, under kind of requirements we set it to be running with, and essentially observe what is causing the problem... And then generally, at what level of efficiency we are in. So those two things that generally I in my ... |
So this assessment is one thing, and you have to kind of measure that. And that generally, in kind of observability space, we are going into. You have to measure it while running; either running on production, or locally, under your benchmark, we call it. So test, and experiment... |
So assessment is one point. And the second point is bottleneck analysis. So once you know that things are too slow, so you generally know you are, for example, one second behind your expectations of the HTTP request, for example, you can go and find some tooling around observability that will tell you the bottleneck. S... |
**Mat Ryer:** Does anyone disagree with that? |
**Kemal Akkoyun:** Everyone agrees. Everyone works on observability and monitoring, so who can disagree...? |
**Mat Ryer:** Yeah, exactly. Good point. Okay, so observability then... So this is literally like monitoring resource usage, or the time it takes to do things... Is there a limit, really? How do we know what to measure? |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.