text stringlengths 0 2.35k |
|---|
**Ole Bulbuk:** Yeah. I mean, at the start it was quite exciting. We tried to keep it all running and do everything exactly the same as before. After a while we saw that it's not that exciting anymore, and people start to get Zoom fatigue, and whatever. Members were a bit decreasing... And then we saw that we got more ... |
**Natalie Pistunovich:** For sure. And we had some attempts of saying "Maybe next month we'll try" or "Maybe in spring we'll try", but then always numbers were against us, and regulations and whatnot. You mentioned that we had a decrease in attendance, but changed the character to be more international; that's definite... |
In the past we met once a month always. Since forever it was always once a month, but it would move around when in the month. Then at some point we anchored the second Wednesday of the month and we started recording, and then uploading this to YouTube... So less attendance, but this instead kind of compensates with mor... |
We do look to get back in-person, maybe this spring/summer, which is exciting... And it will change things again. So we'll see if the Zoom fatigue is being replaced with "I'm used to not leaving the house. I'm not gonna come to events" fatigue. Maybe yes, maybe not. Definitely, a global thing to think about, which is m... |
**Jingle:** \[53:39\] to \[53:58\] |
**Natalie Pistunovich:** Ole, I heard a rumor, that you have two unpopular opinions. |
**Ole Bulbuk:** Yeah, that is true. The first is a bit of a meta opinion. My first unpopular opinion is that I think the popularity of an opinion depends more on the audience than on the opinion itself. A very good example would be stating that Go is a low-latency, low-overhead programming language. When you tell this ... |
So this is something quite different... And I think the audience is a major factor. |
**Natalie Pistunovich:** Okay, so in order to be ranked high in the Unpopular Opinion list of Go Time, you want to optimize for something that is very unpopular with Go developers. |
**Ole Bulbuk:** So I could now state an opinion like "Go is a very slow, large latency and whatever language." I don't think I will find an opinion like that that it's -- I'm a gopher myself, come on... \[laughs\] Maybe I have a second opinion that it's a bit more broader also... |
**Natalie Pistunovich:** Is that optimized for the Go crowd? |
**Ole Bulbuk:** No, it's not. I think it's optimized more in general for the IT crowd. And it's more about how universities and so on work. Because they all do something very nice, abstract, and a lot of algorithms and data structures, or how to create programming languages, and how to study them, or how to work and im... |
I think a lot of the over-engineering we see in the real world stems from the discrepancy that we see between the problems we learn to solve at universities and the problems that we have in real life, or lack of problems maybe. |
**Natalie Pistunovich:** \[57:05\] Interesting. I will not lie, I feel a little bit attacked... |
**Ole Bulbuk:** Sorry. \[laughter\] |
**Natalie Pistunovich:** Over-engineering is definitely a topic I have heard in the past... That sure is interesting. I agree with you that, unfortunately, this is not so unpopular, at least in our little sample group. I agree with you that the type of skills and the views on programming that you get in universities is... |
It's still hard for me to sometimes let go of doing the right thing versus doing the thing that is more efficient on the business part. I have sinned in that multiple times in the past, and it's still ongoing. Maybe if I would have learned more in university, more programming in the context of "This is the right thing,... |
**Ole Bulbuk:** Well, sorry... \[laughter\] |
**Natalie Pistunovich:** But we'll have a survey on Twitter and we'll see how many people agree. |
**Ole Bulbuk:** Okay. |
**Natalie Pistunovich:** Yeah, fun. Thanks a lot for joining. We talked about so many different interesting things, and we'll definitely include in the show notes also all the things that we mentioned and are relevant. And join the meetup, join our user group if you wanna see Ole and me. |
**Ole Bulbuk:** Yeah, please do. |
**Natalie Pistunovich:** Thank you, Ole. |
**Ole Bulbuk:** You're welcome. |
**Natalie Pistunovich:** Have a good evening. |
**Ole Bulbuk:** Bye. |
• Logging as a ubiquitous practice in software development |
• Debate on whether to remove logging statements after initial debugging phase or keep them for future reference |
• Importance of context in log messages, including machine information, user IDs, and trace IDs |
• Benefits of standardizing log format for easier analysis and correlation across systems |
• Trade-off between verbosity and relevance in logging, with some arguing that excessive detail is unnecessary |
• The importance of structured logging for machine readability |
• Comparison between JSON and logfmt (a key-value pair format) for logging |
• Discussion on the trade-off between human readability and machine parsability |
• Recommendation to keep a flat structure in logs, especially with JSON |
• Consideration of what belongs in logs versus what belongs in a database |
• Definition of structured logging as opposed to event logs or access logs |
• Advice against storing primary application logs in the same system that needs to run them (e.g. not storing logs in the database) |
• Consistency of log output with key-value pairs in logfmt format |
• Contextual information in logs for easier pattern recognition |
• Use of context to carry contextual information such as user ID and hostname |
• Challenges with using context deadline exceeded errors in distributed systems |
• Difficulty in distinguishing between different types of context cancellation |
• Potential improvements to Go's error handling, including adding a string parameter to the cancel function |
• Error messages should be unique within an app |
• Logs can be used for error handling and troubleshooting |
• Including context in logs can help with debugging complex systems |
• Writing log entries for the audience, not just for oneself |
• Centralizing error strings for easier maintenance and internationalization |
• Log levels (debug, info, warning, error, critical) and their use cases |
• Use of separate packages for developer vs production logging |
• Benefits of having runtime log-level changing capabilities |
• Importance of a standardized interface for logging in Go |
• Trade-offs between log verbosity, allocation rate, and performance impact on applications |
• Mat Ryer quizzes Jon Calhoun on Java's println methods |
• Discussion of logging vs metrics and the trade-offs between them |
• Ed Welch explains Loki as a time-series database for strings |
• Benefits and drawbacks of combining logs and metrics in one system |
• Importance of specialized tooling (logs, metrics, traces) for big distributed systems |
• Use cases for including assertions about logged messages in testing |
• Logging vs metrics: different approaches to software development and what to prioritize |
• Importance of event timestamp accuracy in logs versus metrics |
• Challenges of dealing with large amounts of log data (petabytes) |
• Loki's approach to indexing metadata instead of full text, and its optimization for parallelism and object stores |
• Unpopular opinions segment: Ed Welch mentions he doesn't have an unpopular opinion but laughs about the goal of having one |
• Integration testing being a net loss |
• Ed Welch's unpopular opinion on not doing integration testing |
• Difficulty with large-scale integration tests, including false positives and maintenance issues |
• Value in having integration tests available for local development, but not as a hard requirement |
• Running integration tests against operational data or clusters |
• Keeping integration test scope small and purposeful, running them on-demand |
• Ed Welch's opinion that Windows is the best desktop OS |
• Comparison of Windows to macOS and Linux |
• Keyboard shortcuts and copy-paste functionality differences between Mac and Windows |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.