| • What is GraphQL
|
| • Solution to classic difficulties of REST APIs (query language for fetching data and spec for implementing GraphQL servers)
|
| • Nested queries and fields (client can select specific fields needed)
|
| • Normalizing data (client doesn't have to select field if not needed)
|
| • Server-side work vs client-side work in GraphQL
|
| • Comparison with REST APIs (baked-in nested resources, query complexity)
|
| • GraphQL's complexity on the client-side is manageable with libraries like Mat Ryer's Machine Box GraphQL client library
|
| • Server-side implementation of GraphQL can be complex due to resolving data and handling nested fields
|
| • gqlgen is a server library that helps generate resolver functions for each field in the schema
|
| • Client libraries can simplify query generation, such as shurcooL's GraphQL client which uses struct tags to generate queries
|
| • GraphQL allows for optional resolvers and reduces complexity on the client-side by allowing clients to request only the necessary data
|
| • Server implementation of GraphQL requires handling database interactions, including joins in SQL databases or document databases
|
| • GraphQL is beneficial for solving problems related to API design, such as reducing the number of requests needed from clients and making it easier for frontend developers to modify queries.
|
| • Problem of deciding what data to load when using a public API
|
| • Comparison between GraphQL and SQL for simplicity and complexity
|
| • Discussion of empowering clients with flexibility in querying data
|
| • Mention of alternative approaches like Remix and static query rolling
|
| • Trade-offs between client empowerment and potential performance losses
|
| • Strategies for optimizing queries and protecting against denial of service
|
| • Discussion of GraphQL features and benefits
|
| • Complexity approach for public APIs with complexity limits instead of rate limits
|
| • Federation in GraphQL and its implementation
|
| • Data structure impact on query performance
|
| • Schema-first libraries like gqlgen and their benefits
|
| • Importance of API design and documentation
|
| • Comparison between GraphQL and JSON APIs
|
| • GraphQL is a typed language that encourages designing schemas to clearly define data types
|
| • The challenge of matching client and server types was overcome by writing a code generator (genclient) that creates correct Go types from the schema
|
| • Having both server and client use type systems allows for better collaboration and error checking across the stack
|
| • GraphQL's benefits extend beyond just small projects, including improved confidence in API design and evolution
|
| • The relationships and traversability capabilities of GraphQL are particularly useful for modeling complex data structures
|
| • When to start using GraphQL: wait until REST becomes frustrating due to complexity and scalability issues
|
| • Benefits of GraphQL: simplifies data retrieval, reduces complexity, and improves query flexibility
|
| • Relational data: most data has relationships between entities, making GraphQL valuable for querying related data
|
| • Public APIs: releasing a GraphQL API may deter some developers due to learning curve and increased cognitive effort
|
| • Federation: a method for connecting multiple services and allowing them to work together seamlessly
|
| • Automation: manually doing tasks first can lead to better understanding of the problem and more effective automation later on
|
| • Developers often make the mistake of trying to automate tasks unnecessarily
|
| • Automation can be part of a solution, but not always the best approach for every task
|
| • Union types in programming languages would be a useful feature
|
| • Running fiber internet is ideal, but may not be feasible for everyone
|
| • ISPs and rural internet infrastructure can be unreliable and frustrating
|
| • The hosts discuss the topic of internet speed, with Mat Ryer joking that a slow connection is only possible if people use it.
|
| • Jon Calhoun shares an anecdote about his uncle who worked at Comcast and joked about the internet getting slower as users moved back to a particular town.
|
| • The idea of "phoning someone's uncle" for an interview or discussion segment is introduced, with Mat Ryer suggesting they do this in a future episode.
|
| • The conversation touches on the potential for uncles to have unpopular opinions and interesting views. |
