SOLID Go Design_summary.txt

• Introduction to Dave Cheney and his background
• How Dave got started with writing about Go on his blog and its impact
• Dave's contributions to the Go project, including hosting Arm builds and proposing language changes
• Dave's experiences as a developer at Canonical and traveling for Go conferences
• Gratitude from the hosts and guests for Dave's efforts in promoting Go and building community
• Discussion on the importance of design principles for long-term maintainability of Go code
• Critique of the focus on "good code" as being subjective and not actionable
• Proposal for a more objective approach to design using guidelines rather than rules
• Value of discussing design at an abstract level, focusing on goals rather than specific patterns or solutions
• Maturity model for Go, including potential growth phases and lessons from other programming languages
• Discussion about the Gang of Four book and its influence on software design patterns
• Limited number of fundamental software design patterns, with 30-odd being considered sufficient for most scenarios
• Comparison to laws of nature, implying a finite set of underlying principles
• Debate on algorithmic complexity and trade-offs (time vs. space)
• Meta-language for discussing algorithms (big O notation, time and complexity)
• Design decisions in software development, including coupling, lookups, and package layout
• Critique of the standard library as an example of inconsistent design
• Evolving knowledge and code design over time
• Discussion on error handling and a new approach being advocated by Dave Cheney
• Evolution of functional options in Go
• Error handling design: fail-fast, fail early
• Importance of decoupling and simplicity in error handling
• Use of interfaces for modular design and loose coupling
• Considerations for retrying operations and idempotency
• Information hiding and encoding extra information into errors
• Sticking additional context to errors using fmt.Errorf
• The standard library in Go has a pattern of returning errors with descriptive messages
• Checking for specific error values can be problematic and lead to issues with stacking errors
• A proposed solution is to give errors a method that allows getting the underlying error and undoing stacking
• Using sentinel error values based on type can become problematic
• Tagged logs only help in log messages, not when passed back up the stack
• Handling errors once at each level of the call stack can lead to excessive logging
• Proposed solution is to return the error with annotations to the caller and handle it there
• Structured logging is seen as unnecessary for operator use cases, but useful for developers during development.
• Different personas for logging (developers vs operators)
• Structured logging and its limitations
• Use cases for counters and metrics instead of logs
• Distributed tracing and request IDs
• Ordered logs and their importance
• Instrumentation and monitoring versus logging
• Trade-offs between logging, performance, and storage costs
• Go's approach to error handling is a key factor in its success for writing server software
• Error handling in Go does not use exceptions but rather requires explicit checks for errors
• The use of error handling in Go encourages developers to think about potential failures and handle them at the point of failure
• The "errors" package can simplify error handling by allowing returns of error values with nil indicating no error
• The verbose nature of error handling in Go is a design decision that prioritizes reliability over convenience
• There are parallels between designing interfaces in Go and error handling, both require thinking about potential failures and handling them at the point of failure
• A lack of clear guidance on when to use channels and how to structure concurrent code is an open question in the Go community
• There is a growing interest in discussing language design and best practices for writing successful Go code.
• The hosts discuss their time constraints and decide to skip over certain topics
• Brian Ketelsen talks about his experience with rsync, a UNIX tool for synchronizing files
• Dave Cheney mentions the connection between Samba and rsync, and recommends pt (Platinum Searcher) as a faster search alternative to Ack or AG
• Carlisia Thompson shares her experience using Sourcegraph, which she finds much faster than grep
• Erik St. Martin talks about Asciidoctor, a tool for generating documentation with features like table of contents and source code highlighting
• The hosts also discuss their personal preferences for text editors and search tools
• Show submission and guest suggestions via GitHub
• Wrap-up and goodbye