text
stringlengths
0
2.35k
**Mat Ryer:** Yeah, you can't polish a turd, is what you're saying. Normally, we challenge them, you know...
**Ed Welch:** Shine it up a little bit though...
**Mat Ryer:** Yeah, that's my unpopular opinion - I actually think... Yeah, you probably can polish a turd, to be honest.
**Ed Welch:** The Myth Busters actually did.
**Mat Ryer:** And they've polished a turd on Myth Busters?
**Ed Welch:** Yeah.
**Mat Ryer:** Okay. We've said turd too many times for a family-friendly podcast.
**Jon Calhoun:** We start with the potty humor and end with the potty humor.
**Mat Ryer:** We did. It's a potty sandwich.
**Ed Welch:** Is this still on the show? Are we still going?
**Jon Calhoun:** I mean, it's recording right now...
**Mat Ryer:** Yeah, yeah, we're still live. I haven't played the final -- this is all legally binding until I play the final music, so...
**Ed Welch:** Gotcha.
**Mat Ryer:** Well, we'll put that in the show notes, the Myth Busters thing, if this makes it in... Well, thank you so much again, and thanks for listening. See you next time, on Go Time!
• Definition of API: discussion of different perspectives on what an API is, including a contract between services and a series of endpoints providing resources.
• APIToolkit project: goal of detecting anomalies in APIs before customers find them, using machine learning to understand usage patterns and potential impacts of changes.
• Operability of APIs: challenges of making changes to public APIs after they are published, need for strategies to manage versioning and backwards-incompatible changes.
• Anomaly detection: APIToolkit's approach to identifying issues in API usage, including detecting changes in field formats or values that may indicate a problem.
• API versioning and compatibility issues
• Importance of testing and monitoring APIs
• Role of tools like APIToolkit in detecting changes and issues
• Accountability and tracking changes to API contracts
• Observability and monitoring APIs, including field-level statistics
• Use of Go programming language in the APIToolkit project
• Use of Go and Haskell languages to optimize performance
• APIToolkit's function as middleware, processing requests and sending sensitive data to servers
• Client-side integration of APIToolkit through language-specific middleware or sidecars
• Long-term goal is a collector-based system inspired by OpenTelemetry for efficient data collection
• Current implementation is mostly Go with some Haskell usage for specific tasks
• Returning HTTP 200 instead of error codes for errors
• Sending JSON bodies as strings or with custom error codes
• Third-party provider issues with inconsistent field handling (null vs empty string)
• Handling JSON in Go and potential challenges due to static typing
• Processing JSON at different stages and its impact on performance
• APIToolkit is an open-source project on GitHub, but currently in a closed beta for select companies
• The project needs to handle high traffic and requests per second, requiring flexible time constraints
• Sampling may be necessary for large-scale testing, but not always required
• Community assistance would be helpful in developing clients for various programming languages
• APIToolkit's core middleware is available on GitHub, with a waitlist for early adopters
• The current school system may not be effective, with excessive homework and long hours potentially detrimental to children's well-being
• Alternative education methods, such as homeschooling, have shown better results in some studies
• Anthony Alaribe shares his positive opinion on the German language after learning it.
• He finds the rules of German clear and logical compared to English.
• The group discusses the stereotype that German is a harsh language with many exceptions.
• The conversation ends with a humorous comparison of languages as "API contracts" between individuals and governments.
**Natalie Pistunovich:** We are live on the tubes, and welcome everyone, from whatever hour of the day you're joining us, and whenever you're hearing this, live or recorded. I think for once this is a European majority here, so we can say good evening...
I'm joined today by my co-host, Johnny, and we have two guests. We have Anthony and we have Smile. Anthony, you are a senior software engineer at Delivery Hero here in Berlin, and you've been working with Go since 2013, and you're a co-founder of APIToolkit. And Smile is your co-founder, and you are a senior mobile eng...
**Anthony Alaribe:** Thanks for having us, Natalie.
**Smile Egbai:** Thanks. Nice to be here.
**Natalie Pistunovich:** Okay, so what does API stand for? Let's ask the big questions first. \[laughter\]
**Anthony Alaribe:** Huge question. No pressure.
**Natalie Pistunovich:** No pressure. I had to google that before coming into the episode... So anybody who's listening and googling right now, we get you.
**Anthony Alaribe:** I mean, an API is an application programming interface, but I think it can mean a lot of things. I like to think of an API as a contract. So if you have, for example, two services, two machines, one server and one mobile application and they need to communicate with each other, both parties can dec...
**Natalie Pistunovich:** \[04:13\] Okay. Smile, what is your take on what is an API.
**Smile Egbai:** My take on what an API is roughly the same thing, and coming majorly from the mobile background, I would always say an API is more or less a series of endpoints that will give you resources.
**Natalie Pistunovich:** A series of endpoints that will give you resources. That's short and to the point.
**Johnny Boursiquot:** I like those definitions, they're very specific in terms of use. That's one way to use an API. Maybe you're making some HTTP requests, maybe you're dealing with a different transport mechanism, maybe you're doing it through gRPC, maybe you're dealing - perhaps fortunate or unfortunate - with some...
**Anthony Alaribe:** Hopefully not...
**Johnny Boursiquot:** \[laughs\] You know, a lot of people still make a living on it, so I'm not knocking it. But to me, these are ways - how do you communicate with an API. To me, an API is, as the name implies, an Application Programming Interface, it's the interface to your functionality, to your code, to the behav...
**Smile Egbai:** Yeah.
**Anthony Alaribe:** Even if you have a codebase, maybe just like a library or a Go package, like you mentioned, with a series of exposed functions, those exposed functions are also part of the API.
**Johnny Boursiquot:** And building the APIToolkit project - I'm very curious about what the offering is. With a name like APIToolkit, it sounds like you're sort of bringing something generic, in the sense of something that is general-purpose rather, for APIs. Did I get that about right? What is it that you're bringing...
**Anthony Alaribe:** You know the two most difficult problems in computer science - naming things and the rest... \[laughter\] Yeah, finding names is hard, but I think this name kind of sums up a lot of what we're trying to do. We're trying to solve a few problems around APIs, and the name kind of gives it room to grow...
The main problems that APIToolkit is trying to solve is around anomaly detection. So you have an API, in this case a primarily RESTful API, so an API that you expose over a transport. But let's stick with a REST API. So you have a REST API, with a series of endpoints, and each endpoint returns a series of fields, with ...
\[08:01\] Or even for example you have a legacy service which you're trying to rewrite, for different reasons, and you believe you know about the contract of that API, because it was very well documented... But then you rewrite that API and realize that - okay, there were some fields were there maybe ten years ago, no ...
**Johnny Boursiquot:** So that's an interesting problem space... So if we can start to sort of dig in a little bit here - when I think of the operability of an API - say you make something public, and once you make it public, like you said, you basically have a contract with whoever is gonna be consuming that. So makin...