Chris Benson: I think there’s certainly a chance at it, and I would argue – it’s the same argument I’ve made in previous shows where we talked on similar topics, is that we’re seeing… As the AI industry is has been maturing these years at an incredibly rapid pace, but we’re still seeing many of the things occurring that we saw when the software world was really maturing over several decades. And the place where open source has really, really worked are in common touchpoints where all organizations or many organizations need a common thing. And they might build something differentiated on top of that, for their revenue, to drive profitability… But there’s so much that is underneath that point of differentiation that they and many other organizations can get the benefit out of a lot of effort, a lot of work; a lot of times they’ll have paid employees do it…

So there’s a point where working together and doing open stuff makes sense for business, and it drives profitability. It may not be your single point of differentiation, but if it’s anything under that, why not? Why not share the costs, and pool expertise for the best possible foundations? And so what I’m hoping is that we continue to see that play out in the AI space. We’re seeing – if you look at Hugging Face, we’ve already talked about the fact that a couple of months ago they announced that they were hosting a million models; those are all open source. Really, really impressive. And so I think that there is a good chance that a vibrant, open community around AI can and will continue, and it will have a lot of corporate players involved in it. So I’m very optimistic in that way.

Break: [00:13:19.26]

Adam Stacoviak: Not because he like literally loves Jira. It’s like, “This is the best tool for an enterprise like ours. It’s so flexible. It’s so sturdy. It’s got so many APIs and so many ways in and out of it that you could just get things done. So many integrations.” I’m like, “Great. Those are great reasons to love it.” Those are the reasons I don’t love it, because I don’t have those needs, you know?

Lea Anthony: That’s alright. Where was I? Oh yeah, so the new manager came in and said, “Well, what are you using currently?” and I think they were using Postgres, or something. And he said, “Well, for this use case, let’s build it in Postgres.” And the engineers were like “What are you talking about? That’s crazy.” But they did. And the reason why it was such a good idea was because it was a technology they fully understood, they had all the controls in place, so that if something went wrong, they could deal with it. Based on their use case, it was completely adequate enough. It was more than they need it. So whilst it’s not the best tool for the job, it was the best tool for the job.

Autumn Nash: I think that’s the most frustrating part about being an engineer. There’s so many things that will teach you the concepts of writing code, or the concepts of doing general things, but there’s so many more nuances and contexts that you need to actually write production code at scale… And we do a horrible job at teaching people that, and just giving people the documentation. And sometimes you run through documentation and you’re like “Did you even do this to see if it worked before you released this on the ,internet and told people this was going to be how people use your product?” Some of the most successful databases are successful because they have good documentation and it’s easy, not necessarily because it’s the best tool for it. You know?

Kevin Ball: Yeah. I mean, I think debates, for all that they’re uncomfortable, are probably the best tool we have for moving forward our thinking as a community about things. So it’s well worth doing. I think where they start to get in danger is when they happen as clickbait dunks on X or other things like that, where it’s –

Eric Boyd: [01:31:54.20] Exactly. And I was like “I wonder how long using GPT 4 it would take me to write snake in a programming language I don’t know?” And so I chose Go, because I don’t know Go. And in a half hour I had working code. And running, and with graphics libraries, and all that… You write the main loop of the body of the snake, and go. Boom. Here’s the main loop. And I’ll read through it, and like, I’m still a developer, I’ve got to read the code… And I’m like “I don’t understand what you did in this update function.” It seemed to be just truncated. It just made a mistake, it was truncating the snake always the same length… I was like “Shouldn’t the snake grow every time it eats something?” “Oh, you’re right. Here’s the new code for that.” And this back and forth, like I’d have a conversation with an excellent developer, and then it just gave me code that worked in a half hour.

So I think that mental exercise - that’s actually one I’ve asked a lot of people on my team to go do, because it is a new tool, and you kind of have to learn how to use it. When I write code, what do I do? I sit down and I just start typing, and I don’t ask someone “Could you write the main body of this thing for me?” And I think even as we think about emails and documents… Like, if I get a Word doc sent to me, I usually just read it. But maybe I should start asking it, “Hey, could you give me a list of the frequently asked questions from this document?” That’s a really great prompt to give on any document that you haven’t gotten… You get some long email thread, “Could you summarize this for me?” And just sort of learning those habits teaches you to be so much more productive.

And so that’s where I say – I think the productivity potential of this is really incredible… And so if we want to take a little bit sort of the macroeconomic view, world GDP grows because of population or productivity. Population’s flattening, so it’s gotta be productivity. And this is the best tool for productivity growth that I think we have.

Jerod Santo: Because that’s how we choose software a lot of times. Isn’t it emotional? I mean, I feel like it’s so emotional. Are you utilitarian, like “The best tool for the job” every time? Analyze them, break them down?

Pavel Kozlowski: I just want to say that that’s the evolution of the framework; it will take time as we’ll go through different layers of this onion… But regardless of the stage on which we’re in, I think what matters is your applications and your use cases. So we always want to hear about what’s not working. And GitHub issues is probably is the best, the most accessible forum. But just essentially, I think we want to think about all the developers as our colleagues. We work on the same web space. So yeah, let’s just work together on making the best tool possible.

Amal Hussein: Yeah, that’s a great summary. And you’re just reinforcing this notion for me that it’s okay to have different parts of your app be architected differently, right? Optimize for each experience using the best tool for the job; it doesn’t have to be one size fits all.

Justin Searls: That take has a theoretical purity to it that makes me extremely skeptical… [laughter] Because the truth is, sure, there is probably a thing that I can uniquely do, or my team can uniquely do, or that perfectly encapsulates what our project or our application is designed to specialize in. But the truth is that to know where that line ends, and where are the sort of just “This is commodity digital pipe-fitting between two different integrations, or something” begins - that takes time, and it’s probably a moving target. And to know and to be able to discover “What’s the best tool for the job?”, that’s real work, and keep all that stuff upgraded as real work. And so I think a lot of the early npm employees were sort of “package small things” extremists, and I thought that I was one, and then they really like lapped me… And it can go so far that you’re spending all of your time just trying to curate and keep up with this gargantuan pyramid of transitive dependencies that you’re standing on top of.

Jerod Santo: One-time Lineman.js purveyor, Justin Searls, reflects on the tool’s 10th anniversary:

“Lineman was our attempt to create a pleasant developer experience for building single-page JavaScript applications without tying developers to any particular backend language or framework… I’m biased, but I really believe Lineman was the best tool available at the time for what is now an incredibly common job.”

He goes on to ask himself (and try to answer) a question many of us ask ourselves often: “When I noticed the upcoming anniversary in my calendar, I could only think of one question to reflect on: why did Lineman fail and what can we learn from it today?”

I don’t know if Justin’s analysis is correct or not, but it’s an interesting read. Regardless, he throws in a “how to pick the best tool for the job” section near the end of the post that is worth the price of admission and then some.

Johnny Boursiquot: I don’t know, Kris is a linguist, so he’s gonna slice hairs here… [laughs] He’s gonna split hairs here. But for me, they’re kind of the same thing. I mean, right now I’m in the middle of doing some prototyping, aka proof of concept, aka whatever kind of work, whereby I’m using Go to integrate and to try and do an experiment with a third party system to basically explore the edges of it, see whether it’s going to suit my needs or not. So to me, that is prototyping. So I’m not learning a programming language, I’m learning a system; I’m learning an API, I’m learning a product in this case. So that means I need to build something. So I have a very specific set of things that I’m trying to accomplish, and I use the best tool I know how to sort of experiment. In that case - I use Go every day, so to me, having to define types before I get to use them, I do it so often that it doesn’t slow me, it doesn’t hinder me in any way. If I was learning Go, then I’d have to spend a bit more time trying to understand the type system, trying to understand the quirks, the do’s and don’ts and all that stuff, which to me is different. So basically I’m going on site to do a job, and I’m bringing my tools; I’m bringing my hammer, I’m bringing my saws, I’m bringing my whatever, right? Like, I’m not learning how to use a saw, or I’m not learning how to use a hammer on the job. I’m at a job site, trying to accomplish something, if that analogy makes sense.

Jerod Santo: Check out the Astro website. There’s definitely some cool testimonials for Ruff here. Nick Schrock, who was one of the guys who started GraphQL, called it a game changer. And he’s like “Why? Primarily because it’s nearly 1,000 times faster, literally. Not a typo.” So that aligns with what you’re talking about there, where it’s just so fast that it’s just like why not.

I think it’s cool that… Sometimes you can get – I think I’ve used the word “entrenched” too many times on the show, but you can kind of get like stuck in your own world, and think everything has to be in Python. Python this, Python that, and it’s like, that’s kind of goes against the best tool for the job principle… Because it’s a general purpose tool, but that doesn’t mean it’s necessarily the best for every job. And so if you have a job that you need done, and Rust is way better at it, 1,000 times, in terms of just execution speed on this exact same task, I think it’d be foolish not to look over there and say, “Hey, what if we just used both these tools together, and had a better life?”

Jerod Santo: Every week on Go Time, the panel and their guests share opinions they believe will be unpopular with the listening audience. We take those opinions and conduct Very Scientific Twitter / Mastodon polls to see just how (un)popular they are. Here’s some recent results:

On episode 266, David Wickes laid out a conspiracy theory about singe-page apps. He surmises they were invented and popularized so large corporations like Google can track us. That was 53% popular on Twitter but 60% unpopular on Mastodon.

Also on episode 266, Carson Gross said he thinks too many tech decisions are driven by people’s “fear of looking dumb”. 94% of Twitter pollsters agreed and 91% agreed on Mastodon.

On episode 267, Carl Johnson claimed only barbarians eat salad with a fork. You should use chopsticks instead. This was highly unpopular with 88% disagreeing on Twitter and 84% on Mastodon. Despite this very bad opinion by Carl, it did prompt Johnny to give some very good advice: use chopsticks to eat cheetos while you’re coding.

On episode 271 Lea Anthony shared that he believes the best tool for the job isn’t always the best tool for the job. 95% of poll takers agree with him despite the statement being false as defined.

and last but not least unpopular, even though I think the premise is preposterous, Matthew Boyle said on episode 273 that we should be able to take our laptops with us to the movies. This was 65% unpopular on Twitter, 89% unpopular on Mastodon, and made Mat Ryer respond with…

Travis Fischer: [29:57] Thank you. I view agents as this new – it’s like, if large language models or CPUs and kind of this new compute paradigm, there are these reasoning engines… Yeah, they’re great at generating text, but the real emergent property, the real game-changing property of them is reasoning. If they’re kind of the new reason engines, and you have like, that’s your CPU layer, and then you have a storage layer that’s all these vector databases, and you’re kind of overhyped on that side of things, on top of that you have like “How do you actually run programs?” And that’s agents. And I view it as there’s kind of a spectrum of like traditional programming that might happen to use a large language model, and then on the other end you have like full self-driving agents that are making decisions, and creating tasks, and just fully autonomous. And I’m excited to kind of focus on somewhere in the middle, and focus on more reliable use cases that we can actually build reliably today.

But to your question about kind of the TypeScript/Python world, a lot of the frontier, the libraries at the framework level that are pushing the edge here are all Python-first, right? And I really want to take a TypeScript-first approach; partially because it’s the community that I know and love. It’s my best tool on my tool belt. And partially because I think people building real applications, at the application level, a lot of those folks are more in the JavaScript world.

Adam Argyle: And yeah, a lot of this stuff is – there’s tons of utility in each color space; there’s utility in knowing the JavaScript ways, and the CSS… Just like normal web development, right? You have like this harmony, and a relationship, and an orchestration that you’re gonna find the best tool for the job. And these new color spaces give you better tools for the job. So that’s kind of what’s nice, is after this podcast you’re gonna be like “I have new color tools that help me perform my work better. I like literally learned new functions that give me better results than what I’ve been working with in the past.”

Thomas Eckert: Yeah. All software ships bugs, and you can’t solve all your problems at the exact same time, so you have to have some system to coordinate. And as those systems grow, and you need to report upwards, sometimes moving away from Agile is the right decision.

When it comes to non-agile approaches, you often have to write these documents, that you’re sharing – I mean, you have to read a lot of documentation in an agile environment too, but with waterfall you might have more RFCs, you might have these proposals, these high-level docs… And what I’ve experienced is I’ve not found a perfect tool for writing those docs, getting feedback from multiple stakeholders. I know some teams use Dropbox Paper, I know that we use Google Docs, some teams use Notion… What have been your experiences? What’s the best tool for sharing those non-code, specific ideas and getting feedback?

Lea Anthony: That’s alright. Where was I? Oh yeah, so the new manager came in and said, “Well, what are you using currently?” and I think they were using Postgres, or something. And he said, “Well, for this use case, let’s build it in Postgres.” And the engineers were like “What are you talking about? That’s crazy.” But they did. And the reason why it was such a good idea was because it was a technology they fully understood, they had all the controls in place, so that if something went wrong, they could deal with it. Based on their use case, it was completely adequate enough. It was more than they need it. So whilst it’s not the best tool for the job, it was the best tool for the job. And so that’s an interesting – that’s something interesting I learned recently.

Aaron Schlesinger: [27:53] Yeah? Okay. It’s got the Johnny stamp of approval… So you can turn PostgreSQL into a time-series database, if you want to, for example, right? By using very advanced features of SQL. But not many people do that, because obviously, a lot of times it’s not the best tool for the job. But also, most people prefer, including myself, to stick with a smaller set of foundational features of the technology. And same thing with FP. There is a smaller set than even all the stuff that I said today, there’s a smaller set of functionality that most people prefer to use, and my hypothesis for why that is is because it makes more sense, and it fits into more workloads.

So if you need to capture some state of a function, and expose one operation on that state to the caller, using a higher-order function makes a lot of sense. It’s a lot easier than doing a bunch of boilerplate, and building a whole struct, and storing the state in the struct, and having a bunch of methods on it. And that - it just fits right into that use case, and that use case is applicable and exists on a ton of workloads, and so we use it. So FP, strictly speaking - it is everywhere.

Now, I’m going to modify your second question a little bit, accordingly… You said “Should we be trying to get FP into more workloads?” Well, given that it’s already in a lot of places, I’m going to modify it and say, “Should we get more features of FP into more workloads?” Will you allow that? [laughter]

Emma Draper: Yeah, I’ve had mixed feedback on the different tools. If you’re considering them, definitely do research on it. I know I’ve used some that seemed kind of helpful, and some that have been dropped, because you just don’t get very good signal from them… In a perfect world, I would like to never use them, or have to use them, because ultimately, I think you get the best signal from being in front of a person and having stuff specific to what you’re trying to hire. I do understand their place… If you are really rapidly trying to scale and hire a lot of people, and you don’t have a lot of existing engineers that can spend their entire time interviewing, so I can see the place for it… But ultimately, I think it’s not the best tool, and I think particularly if you’re trying to hire more senior and experienced people, you’re never going to get the signal from that. You might even turn away candidates with stuff like that… So use it wisely, sparingly, know what it’s good for and what it’s not… Have you used them much, Emma?

Jon Buda: It turns out it’s the best tool for everything. It is kind of amazing… I mean, obviously, we rely on FFmpeg, but it’s kind of amazing that all this software is just open source, for us to use… None of this would be possible if all these libraries weren’t in existence, and we couldn’t just use them. So I don’t know, sometimes when I think about that, it’s kind of amazing that we can get away with building a company that provides us a living, built on free software.

Nayana Shetty: Yeah. And also within teams. When there are so many different teams, and every team is autonomous to use their own tools… I think you need the core principles to be the same, irrespective of what tools they’re using, and that’s where I find having the capabilities and the principle layer set right would help teams figure out what is the best tool for their use cases and stuff.

Ben Ford: Yeah. There’s better alternatives. So if you’re using Excel and email, you are like an athlete with the flu. You are not going to win. Even if you would normally win, even if you’ve got impeccable training, if your muscles and your sinews are equivalent of people in a business, they’re really fantastic and you’ve trained them very well… If you’ve trained yourself extremely well, but you have flu, or you have vertigo or something, you’re not going to perform.

It’s exactly the same with businesses. You have to take the best of what’s available out there. And you’re right, Excel is definitely… The problem with Excel, it’s easy to use, so people start using it for databases and for operations, or for managing projects. And it’s not the best tool, it’s not even in the top five of best tools for any of those things. Even for modeling, we’ve got better options like causal.app which I shared with you the other day. So for me, I could just probably return my business, Excel extermination as a service, and it wouldn’t be inaccurate.

Johnny Boursiquot: I mean, clearly, it’s gonna be Go. That’s the language I use. You make me repeat myself. I use it for my day job, I use it in my side projects, I use it for my side-side projects… Go is my go-to. That’s the first thing I think of, even for things that I used to write sort of a quick Bash or a shell script for, or whatever… Or I used to rely on Ruby for a lot of my little scripts, knick-knacks here and there… Because the compiler in Go is so fast, I get that feedback immediately, and I get type safety, and all that stuff… Even the random little scripts that I used to write - I’ve replaced all that stuff with just Go.

So it’s quite possible that I’m now using Go as a hammer to hit every nail, but it makes it so easy for me to do that. But obviously, again, right tool for the job, so depending on what project I’m working on, if there’s already a language that’s being used that’s not Go, I adapt to that. So in my repertoire, I have at least 6-7 different languages, so every once in a while I’ll come across something that “Okay… Man, I haven’t written Lua in ages, but this thing is written in Lua, and now I’ve gotta brush off my Lua.” Or be it Java, or reading Python, whatever it is.

I think the key takeaway here is sometimes you don’t get to pick the best tool for the job… And perhaps that is my unpopular opinion - you don’t always get to pick the best tool for the job. Sometimes the tool that was picked is maybe even the worst possible tool that could have been picked for the job… But that was the tool that was used, so now you have to deal with it. So the unpopular opinion is “Suck it up, buttercup! It is what it is. Sometimes you’ve just gotta do what you’ve gotta do.”

Jerod Santo: There are definitely not as many people who are familiar with Elixir as they would be if it was a Ruby on Rails application, for example, or if it was a JavaScript SPA or something, with a Node backend… But I guess I don’t care all that much about those aspects. I feel like as we build out new areas – I’m not sure I’ve teased it on the show or not, but… I’ve teased it to you, Simey - I’m thinking about doing an API and a Changelog command line, so you can play our shows from your command line, and stuff. And I wouldn’t pick Elixir for those… And that is one of the reasons; I don’t think it’s the best tool for those particular jobs. Maybe for the API, if it’s just part of the app. But for the command line tool, in terms of distribution and stuff, I think Go is a great choice…

[01:08:10.22] And I think that as we have maybe smaller ancillary things, like our transcripts repo, if we had a command line repo that was its own separate project in Go - well, I could see more contributions coming into there, than whatever come against Changelog.com.

Hopefully we open up new opportunities as we move along. I definitely think Elixir as a dependency has probably limited, to a certain degree, outside contributors, but I think more so it’s the lack of vision and clear inroads to contributing that’s done it.

Jerod Santo: “Not the best tool for the job, but we’re going to use it anyways…”

Tammer Saleh: And the sad thing about it is they just don’t know any better. They’ve got very simple applications, they’re a small team, and they end up spending a lot of time building these Helm Charts to distribute them, and stuff. You don’t need that. Customize, for example, is a great tool for managing your YAML when it’s being deployed to multiple environments, because you can make very small changes. Customize is much easier to understand, much easier to maintain. If you’re really small, you don’t even need a tool like that. You could just apply the YAML and just call it a day.

[36:14] I think when a team chooses Kubernetes, where it should focus on is automation; building out their own internal automation system, not just for managing the cluster, using Terraform, which is by far the best tool for that kind of stuff, but also for managing the resources inside the cluster. A CI/CD pipeline, maybe using GitOps at the end, or whatever… That’s the fundamentals that your team should focus on, because once you have that, all the other changes become simpler. And frankly, that automation is half of the value prop of Kubernetes, because the Kubernetes API is so good. It’s so easy to automate stuff through Kubernetes. And if you’re not investing in that automation, you’re wasting that value. And then, obviously – I mean, I run a company, so I should say that… If you’re just choosing Kubernetes, you should be looking for training. And I love our workshops, obviously, but there’s others, right? But you do need to invest in your engineers’ knowledge, because they are going to have to debug it when it goes sideways, and you don’t want them floundering and using Stack Overflow in the middle of an outage.

We offer engineering services, usually not for people who are just now adopting Kubernetes, unless you’ve got a very interesting application you’re moving over, but you should be finding experts, either hiring Kubernetes experts, or finding a partner that you can integrate with your team, that will give you those subject matter experts for Kubernetes, because you’re going to save a lot more time and money in the long run if you do that early on.