Leon Mika

I was going through my drafts this evening and I found that I had a number of them. Around 26 in total, created over the last year or so, many of them half-finished or almost finished thoughts. Some are not worth keeping around, and will be going straight to the big /dev/null in the sky. But I found a few that are worth preserving, at least in some form. And since they’re no longer worth publishing as separate posts in-and-of-themselves — which is probably why they were kept as drafts — I’d thought I’d just list them here.

So here they are:

• Feb 2024: Tried the popover API for the first time today, and was quite impressed as to how easy it was to use. I needed to add a new modal for Photo Bucket¹, and to be able to do so using only HTML is great.
• Feb 2024: B-movie idea: Two mystical titans — Gozilla and Kong — team up to fight the biggest villain of them all: humanity.
• Mar 2024: I read a blog post about travel tips, and the poster recommend “tak[ing] one week trips. Turns out, two weeks is just too long.” This is where the tannery of distance places a role. Being ask to travel half-way round the world for five or six days at a time is not a great feeling. By the time I recover from jet-lag it’s time to head back home again. So I think a week is too short a time for me. Experience has taught me that I can probably go up to three weeks before I start feeling homesick, so I think 2-3 weeks would do it.
• Apr 2024: Forgot how much of a pain configuring Linux audio using the command line actually is. Got there in the end with some old hardware that I know works with Ubuntu.²
• May 2024: I think the secret to sustainable mocking is to avoid trying to litter your tests with hard-coded inputs and outputs, and instead implement mocks which satisfies the contract with the function under test with naive, in-memory implementations of dependent services.³
• Sept 2024: I don’t know how this came about, but there seems like there’s an approach to Go code which shuns writing a lot of little private functions that in favour of big ones that inline all the logic. The problem is that this results in a lot of noise. Go’s naturally a boiler-plate heavy language, and if a large routine contains both the business logic and all the low-level mechanics of, say, trying get a unique list of things in a slice, it’s hard to work out whether that if statement is handling control logic, business logic, or a mixture of the two. It becomes difficult trying to follow what the routine’s trying to accomplish. That’s why I still prefer to write a lot of small methods that break up these large routines. It means a deeper stack and more jumping around a file, which does make it easier to loose your place, but what it gives me is the ability to abstract things out so that the business logic is clearer to others.
• Sept 2024: Sometimes I wonder if I should work towards something that can be a good enough showcase of my professional self that can replace my LinkedIn profile. To have something akin to inessential.com. Not sure what that could be, but I don’t feel like I’ve got that something just yet.
• Sept 2024: I may have found a potential use for GitHub Copilot after all. It’s not to give insight or ask questions about my own code. After all, I like to think I know my own code. No, it’s to provide insight or ask questions about some other open-source project, especially when the documentation is unclear or incomplete.⁴
• Oct 2024: I’m not a blogger, but I play one on the Internet.
• Oct 2024: Upgraded Bike to the latest release which has a caret “bounce” animation when you deselect something, probably to indicate where the carat is at any point. And I absolutely do not like it.⁵
• Feb 2025: How much did the Vic government pay to host the NFL for a showcase game at the MCG? If they paid more than $0.00, they paid too much.
• Feb 2025: Someone at worked asked if I missed working in Java. I don’t think I’d say I hate working in Java. I think the language itself is pretty decent. What got me frustrated working it it was all the non-Java aspects of working on enterprise applications intended for application containers. Many of these technologies involved a lot of configuration — usually XML — that provided no easy way to troubleshoot should applications fail to deploy, an almost near certainly when spinning up a new project.
• Mar 2025: Complaining about excessive pop-overs and chat assistants on websites is a perennial topic, but good grief… do these site owners ever use these sites themselves?
• Mar 2025: It seems to me like people are expecting too much from LLMs. I see take-down posts from people using them to get suggestions on how to do something, mocking the suggestions, then concluding that LLMs lack the nuance decision making that comes from an expert. Yeah, of course they do. They’re statistics-based averaging machines. They’re not following the same thought process that a human will go through.

And that’s it. Hope this wasn’t a significant waste of time.