text stringlengths 0 2.35k |
|---|
[3390.68 --> 3391.34] Your thoughts? |
[3391.86 --> 3393.56] Let us know in the comments. |
[3394.02 --> 3397.50] There's a direct link to the discussion thread at the top of your show notes. |
[3397.50 --> 3400.72] Everyone on this episode will be notified of what you have to say. |
[3400.88 --> 3405.72] So it's a great place for follow-ups, clarifications, links to related projects, stuff like that. |
[3406.04 --> 3407.86] If you're a first-time listener, welcome. |
[3408.34 --> 3410.96] Don't forget to subscribe at gotime.fm. |
[3411.18 --> 3413.12] We are also in your favorite podcast app. |
[3413.30 --> 3414.20] Just search for Gotime. |
[3414.20 --> 3419.02] If this is your 10th listen, your 100th, your 1,000th, whatever order of magnitude you have with us, |
[3419.14 --> 3421.26] we'd love a review and recommendation. |
[3421.82 --> 3425.14] Special thanks to Fastly for CD ending for us all these years, |
[3425.36 --> 3427.26] to Breakmaster Cylinder for the Fresh Beats, |
[3427.50 --> 3428.42] and to you for listening. |
[3428.64 --> 3429.42] We appreciate you. |
[3429.82 --> 3432.34] Next week, Matt and John are joined by Ed Welch |
[3432.34 --> 3434.58] to discuss logging, logging, and more logging. |
[3435.00 --> 3436.24] Yeah, a lot of logging going on there. |
[3436.88 --> 3439.88] That's one to look forward to next time on Gotime. |
[3439.88 --> 3447.68] END |
[3447.68 --> 3447.90] STAN |
[3447.90 --> 3449.44] TO believe in themarkt |
[3449.44 --> 3450.34] STAN |
[3450.34 --> 3450.70] END |
[3450.70 --> 3451.28] Nine |
• Protocol buffers are a language for writing schemas for data and a binary format for serializing data. |
• They were created by Google to simplify and improve efficiency in programming, especially when dealing with APIs and multiple clients. |
• A protocol buffer schema is similar to a Go struct but can be used across languages and formats (e.g. JSON) with a code generator. |
• Using protocol buffers allows programmers to avoid error-prone tasks of rewriting data structures for each client language. |
• Trade-offs include the need for a schema and potential compatibility issues, but especially in Go, these are considered minor compared to the benefits. |
• Protocol buffers can be used as a schema language even when working with JSON. |
• Comparison of Protocol Buffers to Swagger (OpenAPI) for data modeling |
• Critique of JSON schema's verbosity and lack of native types |
• Discussion of the trade-offs between static typing (Protocol Buffers) and dynamic typing (JSON) |
• Comparison of Protocol Buffers to GraphQL |
• Suitability of Protocol Buffers for exposing public APIs, particularly with gRPC |
• Historical challenges with tools for working with Protocol Buffers |
• The complexity of using protocol buffers (protobuf) outside of Google due to its low-level nature and dependencies on other stack components. |
• Benefits of binary formats over text-based ones, including optimized parsing and smaller size. |
• Comparison of protobuf with JSON, including the ability to parse JSON quickly through specialized techniques and compression. |
• Discussion of the Go standard library's encoding mechanisms, including Gob, which is not widely used due to its caveats, such as instability across Go versions. |
• Protobuf's advantages over Gob in terms of performance, stability, and cross-language support. |
• Protocol buffers as a data format for efficient serialization |
• Comparison of protocol buffers with JSON and REST APIs |
• gRPC as an RPC runtime that works with protocol buffers |
• Criticism of gRPC's complexity and verbosity |
• Alternative runtimes such as dRPC and Twirp |
• Discussion on why developers might choose to create their own solutions instead of using existing libraries like gRPC |
• The discussion centers around gRPC, its origins, and its marketing aspect. |
• gRPC is derived from Google's internal system called Stubby, which uses protobuf-flavored HTTP/2. |
• The protocol is relatively simple and could be described in a few paragraphs. |
• The group acknowledges that there are multiple implementations of gRPC, but only one widely-used Go implementation (gRPC-Go). |
• The discussion touches on the idea that having different trade-offs and implementation choices can be beneficial for developers. |
• Buf is working on improving protocol buffers tools, including writing its own protobuf compiler from scratch and a schema registry. |
• gRPC is intimately connected with Protobuf and using one requires considering the other |
• gRPC has strengths in certain areas, but weaknesses in others, such as interoperability with net HTTP |
• Connect is a drop-in replacement for gRPC that is wire-compatible and works with net HTTP |
• Connect aims to make gRPC feel more like a part of the Go ecosystem, rather than a separate entity |
• Perf-wise, using Connect or gRPC is generally comparable to using net HTTP |
• The approach of using statically-typed languages like Go, TypeScript, and Python for APIs is deliberate |
• Even dynamically-typed languages can benefit from type information at boundaries |
• Using gRPC or Connect may be more suitable for internal communications rather than user-facing APIs |
• Developer mindset plays a role in choosing between gRPC/Connect and traditional approaches |
• The benefits of using gRPC with a JSON format for external APIs |
• Limitations of traditional gRPC (e.g., inability to communicate with web browsers) |
• Connect protocol: an alternative to gRPC that supports multiple protocols, including a REST-like one |
• Protocol buffers as a more suitable solution for external APIs due to their flexibility and ability to generate code |
• Importance of being able to call APIs from web browsers and the limitations of traditional gRPC in this regard |
• XML mentioned as an outdated technology and contrasted with modern solutions like protocol buffers |
• Structured logging libraries (such as Zap) being proposed for standard library inclusion |
• Akshay Shah's criticism of structured logging due to performance issues with JSON format |
• Problems with JSON being re-parsed by multiple tools in the log pipeline |
• Preference for binary formats like MessagePack over JSON |
• Jon Calhoun and Johnny Boursiquot sharing their own experiences with logging and structuring code |
• Importance of hands-on coding experience in learning programming |
• The need for showing code and writing it as examples to illustrate concepts |
• Code-alongs encouraged in training programs |
• Launching a new Go course with Lincoln Learning, focusing on hands-on learning |
• Comparison between theoretical understanding and practical implementation of programming concepts |
• Generics in programming and their impact on coding experiences |
• Designing a free course on common algorithms and data structures |
• Teaching without generics initially to focus on core concepts |
• Understanding the role of generics in implementation, particularly with Java's "t" syntax |
• Evolution of Go as a language and its impact on understanding generics |
• Use of generics in code generation for protocol buffers and gRPC |
**Jon Calhoun:** Hello everyone, and welcome to Go Time. Today we're going to be talking about protocol buffers and gRPC, as well as some common misconceptions around them. Today I am joined by Akshay Shah, who works on a profile compiler, schema registry and RPC tools at Buf. Previously, he's worked on several thrift ... |
**Akshay Shah:** I'm good. How are you doing, Jon? |
**Jon Calhoun:** I am doing well. A little bit raspy in the throat, so I apologize for that. I'm also hosting with Johnny Boursiquot. Johnny, how are you? |
**Johnny Boursiquot:** I'm feeling good. I'm tired, but relieved. I've been working on something for a little while, and it's finally coming out tomorrow, so I'm pretty excited about that. |
**Jon Calhoun:** That's awesome. Release days are always fun... Assuming they go well. I hope yours does. |
**Johnny Boursiquot:** Yes... |
**Jon Calhoun:** Alright, so you'll have to let us know about that once that's released. |
**Johnny Boursiquot:** Yeah, indeed. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.